diff --git a/.envrc b/.envrc new file mode 100644 index 0000000000000000000000000000000000000000..3550a30f2de389e537ee40ca5e64a77dc185c79b --- /dev/null +++ b/.envrc @@ -0,0 +1 @@ +use flake diff --git a/.github/cla-signatures.json b/.github/cla-signatures.json index 93c09bf66c3821727f1dd4117766e1b8cd1b6553..0385955af0294ae7f57ebc63661b3d25f9bdd094 100644 --- a/.github/cla-signatures.json +++ b/.github/cla-signatures.json @@ -1615,6 +1615,86 @@ "created_at": "2026-04-23T12:33:57Z", "repoId": 987670088, "pullRequestNo": 2688 + }, + { + "name": "georgeglarson", + "id": 744063, + "comment_id": 4341592963, + "created_at": "2026-04-29T07:08:48Z", + "repoId": 987670088, + "pullRequestNo": 2749 + }, + { + "name": "vorticalbox", + "id": 10886065, + "comment_id": 4344669740, + "created_at": "2026-04-29T14:29:35Z", + "repoId": 987670088, + "pullRequestNo": 2751 + }, + { + "name": "lloydzhou", + "id": 1826685, + "comment_id": 4346011097, + "created_at": "2026-04-29T17:27:37Z", + "repoId": 987670088, + "pullRequestNo": 2752 + }, + { + "name": "carlosgrillet", + "id": 13719979, + "comment_id": 4351020450, + "created_at": "2026-04-30T08:48:52Z", + "repoId": 987670088, + "pullRequestNo": 2759 + }, + { + "name": "SAY-5", + "id": 240962040, + "comment_id": 4357189565, + "created_at": "2026-05-01T01:00:00Z", + "repoId": 987670088, + "pullRequestNo": 2691 + }, + { + "name": "mkaaad", + "id": 119158371, + "comment_id": 4358493210, + "created_at": "2026-05-01T08:25:06Z", + "repoId": 987670088, + "pullRequestNo": 2772 + }, + { + "name": "pragneshbagary", + "id": 41118746, + "comment_id": 4359152162, + "created_at": "2026-05-01T11:54:16Z", + "repoId": 987670088, + "pullRequestNo": 2773 + }, + { + "name": "somjik-api", + "id": 47082582, + "comment_id": 4364741275, + "created_at": "2026-05-02T21:17:55Z", + "repoId": 987670088, + "pullRequestNo": 2783 + }, + { + "name": "ilgax", + "id": 79969769, + "comment_id": 4367211073, + "created_at": "2026-05-03T21:25:46Z", + "repoId": 987670088, + "pullRequestNo": 2790 + }, + { + "name": "ardevd", + "id": 366855, + "comment_id": 4377260996, + "created_at": "2026-05-05T07:22:31Z", + "repoId": 987670088, + "pullRequestNo": 2801 } ] } \ No newline at end of file diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml index 09bdf761361fbb8f0238040a9e2acc0aed3faf57..76206da0173fff2e4dcc5a2afcd308476e260067 100644 --- a/.github/workflows/release.yml +++ b/.github/workflows/release.yml @@ -25,7 +25,6 @@ jobs: fury_token: ${{ secrets.FURY_TOKEN }} nfpm_gpg_key: ${{ secrets.NFPM_GPG_KEY }} nfpm_passphrase: ${{ secrets.NFPM_PASSPHRASE }} - snapcraft_token: ${{ secrets.SNAPCRAFT_TOKEN }} aur_key: ${{ secrets.AUR_KEY }} macos_sign_p12: ${{ secrets.MACOS_SIGN_P12 }} macos_sign_password: ${{ secrets.MACOS_SIGN_PASSWORD }} diff --git a/.github/workflows/security.yml b/.github/workflows/security.yml index 641e21bbb000716579de103f95be9fbdd7d0db8d..5a48dca115ca9cba1ab0810be59db26002791829 100644 --- a/.github/workflows/security.yml +++ b/.github/workflows/security.yml @@ -30,11 +30,11 @@ jobs: - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2 with: persist-credentials: false - - uses: github/codeql-action/init@95e58e9a2cdfd71adc6e0353d5c52f41a045d225 # v4.35.2 + - uses: github/codeql-action/init@e46ed2cbd01164d986452f91f178727624ae40d7 # v4.35.3 with: languages: ${{ matrix.language }} - - uses: github/codeql-action/autobuild@95e58e9a2cdfd71adc6e0353d5c52f41a045d225 # v4.35.2 - - uses: github/codeql-action/analyze@95e58e9a2cdfd71adc6e0353d5c52f41a045d225 # v4.35.2 + - uses: github/codeql-action/autobuild@e46ed2cbd01164d986452f91f178727624ae40d7 # v4.35.3 + - uses: github/codeql-action/analyze@e46ed2cbd01164d986452f91f178727624ae40d7 # v4.35.3 grype: runs-on: ubuntu-latest @@ -52,7 +52,7 @@ jobs: path: "." fail-build: true severity-cutoff: critical - - uses: github/codeql-action/upload-sarif@95e58e9a2cdfd71adc6e0353d5c52f41a045d225 # v4.35.2 + - uses: github/codeql-action/upload-sarif@e46ed2cbd01164d986452f91f178727624ae40d7 # v4.35.3 with: sarif_file: ${{ steps.scan.outputs.sarif }} @@ -73,7 +73,7 @@ jobs: - name: Run govulncheck run: | govulncheck -C . -format sarif ./... > results.sarif - - uses: github/codeql-action/upload-sarif@95e58e9a2cdfd71adc6e0353d5c52f41a045d225 # v4.35.2 + - uses: github/codeql-action/upload-sarif@e46ed2cbd01164d986452f91f178727624ae40d7 # v4.35.3 with: sarif_file: results.sarif diff --git a/.github/workflows/snapshot.yml b/.github/workflows/snapshot.yml index e98a8546fae9cfc1944b1e3682583e15330c8e49..353b1a8d8b4803cb29668ece79272c5c07bd5217 100644 --- a/.github/workflows/snapshot.yml +++ b/.github/workflows/snapshot.yml @@ -25,7 +25,7 @@ jobs: - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0 with: go-version-file: go.mod - - uses: goreleaser/goreleaser-action@e24998b8b67b290c2fa8b7c14fcfa7de2c5c9b8c # v7.1.0 + - uses: goreleaser/goreleaser-action@1a80836c5c9d9e5755a25cb59ec6f45a3b5f41a8 # v7.2.1 with: version: "nightly" distribution: goreleaser-pro diff --git a/.gitignore b/.gitignore index c0bed181bd8e809c3eb461b7b67778272f556911..4c48085beabf0bef48496b03d85693db2df0dd24 100644 --- a/.gitignore +++ b/.gitignore @@ -42,6 +42,9 @@ Thumbs.db .env .env.local +# Nix/direnv +.direnv/ + **/.crush/** /crush diff --git a/docs/hooks/FUTURE.md b/docs/hooks/FUTURE.md index ee334f7308e3264772cbcadc1c581c4e462f937a..bb63185785e6a28c8d0f2c8dda99c93b1845fed4 100644 --- a/docs/hooks/FUTURE.md +++ b/docs/hooks/FUTURE.md @@ -4,7 +4,8 @@ This document tracks planned features and design notes for hooks that are not yet implemented. Nothing here is part of the current contract. Treat it as a scratchpad for what's next, not as documentation of current behavior. -> [!NOTE] This document was largely LLM-generated. +> [!NOTE] +> This document was largely LLM-generated. ## `context_files` diff --git a/docs/hooks/README.md b/docs/hooks/README.md index 1a908a8b644f28170373fdd19908e81a7a92a6a0..30517d3988a1e72d2ebbe093ceac276c3d10400c 100644 --- a/docs/hooks/README.md +++ b/docs/hooks/README.md @@ -1,33 +1,92 @@ # Hooks -> [!NOTE] This document was designed for both humans and agents. +> [!NOTE] +> This document was designed for both humans and agents. -Hooks are user-defined shell commands that fire at specific points during -Crush's execution, giving you deterministic control over an agent's wily -behavior. +Hooks are user-defined shell scripts that run when various events happen during +the agent lifecycle, allowing you to both build on top of Crush, customize +its behavior, and exert deterministic control over an agent's wily behavior. -Hooks in Crush are shell-based, with a focus on simplicity. This allows hooks to -effectively be written in any language. In this document we'll primarily focus -on Bash for simplicity's sake, though we'll include some examples in other -languages at the end, too. +Hooks are just shell commands, and were designed to be both simple and future +forward. ### Hot Hook Facts -- Hooks run before permission checks. If a hook denies a tool call, you'll - never see a permission prompt for it. If a hook explicitly allows a tool - call, you'll _also_ never see a prompt — Crush treats `"decision": "allow"` - as affirmative pre-approval. -- Hooks only fire on the **top-level agent's** tool calls. Sub-agents (the - `agent` task tool, `agentic_fetch`, etc.) run without hook interception so - a single delegated turn doesn't trigger your hook N times. The outer - sub-agent tool call itself _is_ hooked, so policy like "never let the - agent spawn sub-agents" still works. -- Hooks are also compatible with hooks in Claude Code, however this document - covers the Crush-specific API only. One intentional divergence: Crush treats - `updated_input` as a shallow-merge patch rather than a full replacement — see - [Output](#output) below. +- Hooks just shell commands +- Hooks can be written in any language because they’re just executables: Bash, Python, Node, Rust, Haskell, whatever +- Hooks are Claude Code-compatible +- Crush ships with a builtin `crush-hook` skill write, edit, and configure + hooks; just tell Crush how to configure Crush - Crush currently supports just one hook, `PreToolUse`, with plans to support - the full gamut. If there's a hook you'd like to see, let us know. + the full gamut; please let us know which hooks you'd like to see next +- Hooks run in parallel for speed, but their results compose in config order + for determinism + +### Some things you can do with hooks: + +- Block "dangerous" commands: no more `git push -f` or `cabal init` +- Rewrite tool input: turn `node` calls info `deno`, scrub secrets from + commands, rewrite all mentions of "Haskell" into "Haskell, The Best + Language", and so on +- Inject context: add notes to the model's context whenever certain tools are + called. For example: "remember to run gofumpt after editing Go files" +- Auto-approve tools: skip the permission prompt for bash commands that + you know are safe +- Log certain tool calls + +…And lots more. Show us what you're building! + +## Baby's First Hook + +Let's just dive into it and make a simple hook. This particular hook will +disallow the use of Haskell (but we love you, Simon Peyton Jones). + +### Config + +The first thing we need to do is hook up our hook. Let's add the following to +our local `crush.json`. You can, of course, do this globally, too. + +```jsonc +{ + // As expected, hooks go in a "hooks" object. + "hooks": { + // PreToolUse is an event that fires before a tool is used. + "PreToolUse": [ + { + // What tool do we want to hook into? In this case, Bash, because it + // runs the stuff we wanna block. + "matcher": "^bash$", + + // The path to our actual hook script. + "command": "./no-haskell.sh", + }, + ], + }, +} +``` + +Now, let's make our `no-haskell.sh` hook script. + +```bash +#!/usr/bin/env bash + +# Disallow ghc, cabal, and stack. Pipe the bash command output +# ($CRUSH_TOOL_INPUT_COMMAND) to grep and match on a regexp. +if echo "$CRUSH_TOOL_INPUT_COMMAND" | grep -qE '(^| )((ghc|cabal|stack)(\.exe)?)( |$)'; then + + # Someone is trying to use Haskell. Let's send a message back to the model + # and user explaining why we're blocking this. Note that we send all feedback + # like this to stderr. + echo "No Haskell allowed, kiddo." >&2 + + # Now, block the tool call by exiting with code 2. + exit 2 +fi +``` + +That's basically it. For the full guide on how hooks work, however, read on. + +--- ## Execution model @@ -70,8 +129,8 @@ What this means in practice: ## Configuration -Hooks can be added to your `crush.json` at both the global and project-level, -with project level hooks taking precedence. +Hooks can be added to your `crush.json` (or `.crush.json`) at both the global +and project-level, with project level hooks taking precedence. ```jsonc { @@ -87,8 +146,8 @@ with project level hooks taking precedence. } ``` -Hooks are keyed by event name. Only `command` is required; omit `matcher` to -match all tools. +Remember, hooks will run in parallel but resolve in config order. Last hook +wins when rewriting input, but first deny wins when blocking. ## Events @@ -97,29 +156,24 @@ Here are the events you can hook into (spoiler: there's currently just one): ### PreToolUse This hook fires before every tool call. Use it to block dangerous commands, -enforce policies, rewrite tool input, or inject context the model should see. +enforce policies, rewrite tool input, inject context the model should see, log +stuff, and so on. **Matched against**: the tool name (e.g. `bash`, `edit`, `write`, `mcp_github_create_pull_request`). -> [!NOTE] Event names are case insensitive and snake-caseable, so `PreToolUse`, +> [!NOTE] +> Event names are case insensitive and snake-caseable, so `PreToolUse`, > `pretooluse`, `PRETOOLUSE`, `pre_tool_use`, and `PRE_TOOL_USE` all work. -## Baby's First Hook - -Hooks are just shell scripts. Go crazy. +**Scope**: `PreToolUse` only fires on the **top-level agent's** tool calls. +Sub-agents (the `agent` task tool, `agentic_fetch`, etc.) run without hook +interception so a single delegated turn doesn't trigger your hook N times. The +outer sub-agent tool call itself _is_ hooked, so policy like "never let the +agent spawn sub-agents" still works. -```bash -#!/usr/bin/env bash - -# Log all bash tool calls to a file. -printf "%s: %s / %s" \ - "$(date -Iseconds)" \ - "$CRUSH_SESSION_ID" \ - "$CRUSH_TOOL_INPUT_COMMAND" >> ./bash.log -``` - -That's basically it. For the full guide on how hooks work, however, read on. +Hooks are keyed by event name. Only `command` is required, and you can omit +`matcher` to match all tools. ## Building Hooks @@ -133,6 +187,11 @@ When a hook fires, Crush: 4. Waits for all to finish (or time out), then aggregates results **in config order**: deny wins over allow, allow wins over none; `updated_input` patches shallow-merge in order. +5. Applies the result **before** permission checks. If the aggregated decision + is `deny`, the tool call is blocked and you never see a permission prompt + for it. If it's `allow`, Crush treats that as affirmative pre-approval and + also skips the prompt. Silence (no decision) falls through to the normal + permission flow. Note that you can omit `matcher` and match in your shell script instead, however you'll incur some additional overhead as Crush will still parse and @@ -487,6 +546,20 @@ process.stdin.on("end", () => { --- +## Claude Code compatibility + +Crush hooks are broadly compatible with [Claude Code +hooks](https://docs.claude.com/en/docs/claude-code/hooks): the config shape, +stdin payload, output envelope, and exit codes line up so most Claude Code +hooks run under Crush unchanged. This document covers the Crush-specific API +only — anything not documented here isn't guaranteed to work. + +One intentional divergence: Crush treats `updated_input` as a shallow-merge +patch against the original `tool_input` rather than a full replacement. Keys +you omit are preserved. See [Output](#output) for details. + +--- + ## Reference This is the official reference of the narrative above. If prose and this section diff --git a/flake.lock b/flake.lock new file mode 100644 index 0000000000000000000000000000000000000000..88e712ffaf482c6780f1af1f0725831b64ea9cfd --- /dev/null +++ b/flake.lock @@ -0,0 +1,61 @@ +{ + "nodes": { + "flake-utils": { + "inputs": { + "systems": "systems" + }, + "locked": { + "lastModified": 1731533236, + "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=", + "owner": "numtide", + "repo": "flake-utils", + "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b", + "type": "github" + }, + "original": { + "owner": "numtide", + "repo": "flake-utils", + "type": "github" + } + }, + "nixpkgs": { + "locked": { + "lastModified": 1775423009, + "narHash": "sha256-vPKLpjhIVWdDrfiUM8atW6YkIggCEKdSAlJPzzhkQlw=", + "owner": "NixOS", + "repo": "nixpkgs", + "rev": "68d8aa3d661f0e6bd5862291b5bb263b2a6595c9", + "type": "github" + }, + "original": { + "owner": "NixOS", + "ref": "nixos-unstable", + "repo": "nixpkgs", + "type": "github" + } + }, + "root": { + "inputs": { + "flake-utils": "flake-utils", + "nixpkgs": "nixpkgs" + } + }, + "systems": { + "locked": { + "lastModified": 1681028828, + "narHash": "sha256-Vy1rq5AaRuLzOxct8nz4T6wlgyUR7zLU309k9mBC768=", + "owner": "nix-systems", + "repo": "default", + "rev": "da67096a3b9bf56a91d16901293e51ba5b49a27e", + "type": "github" + }, + "original": { + "owner": "nix-systems", + "repo": "default", + "type": "github" + } + } + }, + "root": "root", + "version": 7 +} diff --git a/flake.nix b/flake.nix new file mode 100644 index 0000000000000000000000000000000000000000..c90594ded92d728dea77171a3f5d1b0ac2558c0e --- /dev/null +++ b/flake.nix @@ -0,0 +1,44 @@ +{ + description = "Crush development environment"; + + inputs = { + nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable"; + flake-utils.url = "github:numtide/flake-utils"; + }; + + outputs = { + self, + nixpkgs, + flake-utils, + }: + flake-utils.lib.eachDefaultSystem ( + system: let + pkgs = nixpkgs.legacyPackages.${system}; + in { + devShells.default = pkgs.mkShell { + buildInputs = with pkgs; [ + # Go toolchain + go_1_26 + + # Development tools + gopls # Go language server + golangci-lint # Linter + gofumpt # Formatter (stricter than gofmt) + go-task # Task runner + delve # Go debugger + + # Additional tools + git # Version control + gh # GitHub CLI + svu # Semantic version utility + sqlc # SQL code generator + ]; + + shellHook = '' + # Set Go environment variables + export CGO_ENABLED=0 + ''; + }; + } + ); +} diff --git a/go.mod b/go.mod index 441f37ca687e6cbdbd5faba64777884a374db4a7..cf57b65e28706263e61424f61228568d6ab96704 100644 --- a/go.mod +++ b/go.mod @@ -5,9 +5,9 @@ go 1.26.2 require ( charm.land/bubbles/v2 v2.1.0 charm.land/bubbletea/v2 v2.0.6 - charm.land/catwalk v0.37.14 + charm.land/catwalk v0.39.5 charm.land/fang/v2 v2.0.1 - charm.land/fantasy v0.21.0 + charm.land/fantasy v0.23.0 charm.land/glamour/v2 v2.0.0 charm.land/lipgloss/v2 v2.0.3 charm.land/log/v2 v2.0.0 @@ -16,7 +16,7 @@ require ( github.com/MakeNowJust/heredoc v1.0.0 github.com/Microsoft/go-winio v0.6.2 github.com/PuerkitoBio/goquery v1.12.0 - github.com/alecthomas/chroma/v2 v2.23.1 + github.com/alecthomas/chroma/v2 v2.24.1 github.com/atotto/clipboard v0.1.4 github.com/aymanbagabas/go-nativeclipboard v0.1.3 github.com/aymanbagabas/go-udiff v0.4.1 @@ -43,18 +43,18 @@ require ( github.com/gen2brain/beeep v0.11.2 github.com/go-git/go-git/v5 v5.18.0 github.com/google/uuid v1.6.0 - github.com/invopop/jsonschema v0.13.0 + github.com/invopop/jsonschema v0.14.0 github.com/itchyny/gojq v0.12.19 github.com/joho/godotenv v1.5.1 github.com/jordanella/go-ansi-paintbrush v0.0.0-20240728195301-b7ad996ecf3d github.com/lucasb-eyer/go-colorful v1.4.0 - github.com/mattn/go-isatty v0.0.21 - github.com/modelcontextprotocol/go-sdk v1.5.0 - github.com/ncruces/go-sqlite3 v0.33.3 + github.com/mattn/go-isatty v0.0.22 + github.com/modelcontextprotocol/go-sdk v1.6.0 + github.com/ncruces/go-sqlite3 v0.34.0 github.com/nxadm/tail v1.4.11 github.com/pkg/browser v0.0.0-20240102092130-5ac0b6a4141c - github.com/posthog/posthog-go v1.12.1 - github.com/pressly/goose/v3 v3.27.0 + github.com/posthog/posthog-go v1.12.4 + github.com/pressly/goose/v3 v3.27.1 github.com/qjebbs/go-jsons v1.0.0-alpha.5 github.com/rivo/uniseg v0.4.7 github.com/sahilm/fuzzy v0.1.1 @@ -73,7 +73,7 @@ require ( golang.org/x/text v0.36.0 gopkg.in/natefinch/lumberjack.v2 v2.2.1 gopkg.in/yaml.v3 v3.0.1 - modernc.org/sqlite v1.49.1 + modernc.org/sqlite v1.50.0 mvdan.cc/sh/moreinterp v0.0.0-20250902163504-3cf4fd5717a5 mvdan.cc/sh/v3 v3.13.1 ) @@ -89,7 +89,7 @@ require ( github.com/KyleBanks/depth v1.2.1 // indirect github.com/andybalholm/cascadia v1.3.3 // indirect github.com/aws/aws-sdk-go-v2 v1.41.6 // indirect - github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.8 // indirect + github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.9 // indirect github.com/aws/aws-sdk-go-v2/config v1.32.16 // indirect github.com/aws/aws-sdk-go-v2/credentials v1.19.15 // indirect github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.22 // indirect @@ -102,7 +102,7 @@ require ( github.com/aws/aws-sdk-go-v2/service/sso v1.30.16 // indirect github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.20 // indirect github.com/aws/aws-sdk-go-v2/service/sts v1.42.0 // indirect - github.com/aws/smithy-go v1.25.0 // indirect + github.com/aws/smithy-go v1.25.1 // indirect github.com/aymerick/douceur v0.2.0 // indirect github.com/bahlo/generic-list-go v0.2.0 // indirect github.com/buger/jsonparser v1.1.2 // indirect @@ -113,7 +113,7 @@ require ( github.com/charmbracelet/x/windows v0.2.2 // indirect github.com/cyphar/filepath-securejoin v0.4.1 // indirect github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect - github.com/dlclark/regexp2 v1.11.5 // indirect + github.com/dlclark/regexp2 v1.12.0 // indirect github.com/ebitengine/purego v0.10.0 // indirect github.com/esiqveland/notify v0.13.3 // indirect github.com/felixge/httpsnoop v1.0.4 // indirect @@ -135,7 +135,7 @@ require ( github.com/godbus/dbus/v5 v5.2.2 // indirect github.com/golang/freetype v0.0.0-20170609003504-e2365dfdc4a0 // indirect github.com/google/go-cmp v0.7.0 // indirect - github.com/google/jsonschema-go v0.4.2 // indirect + github.com/google/jsonschema-go v0.4.3 // indirect github.com/google/s2a-go v0.1.9 // indirect github.com/googleapis/enterprise-certificate-proxy v0.3.15 // indirect github.com/googleapis/gax-go/v2 v2.22.0 // indirect @@ -147,10 +147,10 @@ require ( github.com/jackmordaunt/icns/v3 v3.0.1 // indirect github.com/jbenet/go-context v0.0.0-20150711004518-d14ea06fba99 // indirect github.com/josharian/intern v1.0.0 // indirect - github.com/kaptinlin/go-i18n v0.3.0 // indirect - github.com/kaptinlin/jsonpointer v0.4.17 // indirect - github.com/kaptinlin/jsonschema v0.7.7 // indirect - github.com/kaptinlin/messageformat-go v0.4.19 // indirect + github.com/kaptinlin/go-i18n v0.4.4 // indirect + github.com/kaptinlin/jsonpointer v0.4.19 // indirect + github.com/kaptinlin/jsonschema v0.7.11 // indirect + github.com/kaptinlin/messageformat-go v0.6.0 // indirect github.com/klauspost/compress v1.18.5 // indirect github.com/klauspost/cpuid/v2 v2.2.10 // indirect github.com/klauspost/pgzip v1.2.6 // indirect @@ -164,11 +164,12 @@ require ( github.com/muesli/mango-cobra v1.2.0 // indirect github.com/muesli/mango-pflag v0.1.0 // indirect github.com/muesli/roff v0.1.0 // indirect - github.com/ncruces/go-sqlite3-wasm v1.1.1-0.20260409221933-87e4b35a38d0 // indirect + github.com/ncruces/go-sqlite3-wasm/v2 v2.1.35300 // indirect github.com/ncruces/go-strftime v1.0.0 // indirect github.com/ncruces/julianday v1.0.0 // indirect github.com/nfnt/resize v0.0.0-20180221191011-83c6a9932646 // indirect - github.com/pierrec/lz4/v4 v4.1.25 // indirect + github.com/pb33f/ordered-map/v2 v2.3.1 // indirect + github.com/pierrec/lz4/v4 v4.1.26 // indirect github.com/pjbgf/sha1cd v0.3.2 // indirect github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect @@ -184,7 +185,6 @@ require ( github.com/tidwall/pretty v1.2.1 // indirect github.com/u-root/u-root v0.14.1-0.20250807200646-5e7721023dc7 // indirect github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701 // indirect - github.com/wk8/go-ordered-map/v2 v2.1.8 // indirect github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e // indirect github.com/yosida95/uritemplate/v3 v3.0.2 // indirect github.com/yuin/goldmark v1.7.8 // indirect @@ -198,23 +198,23 @@ require ( go.uber.org/multierr v1.11.0 // indirect go.yaml.in/yaml/v4 v4.0.0-rc.3 // indirect golang.org/x/crypto v0.50.0 // indirect - golang.org/x/exp v0.0.0-20260312153236-7ab1446f8b90 // indirect + golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f // indirect golang.org/x/image v0.38.0 // indirect - golang.org/x/mod v0.34.0 // indirect + golang.org/x/mod v0.35.0 // indirect golang.org/x/oauth2 v0.36.0 // indirect golang.org/x/term v0.42.0 // indirect golang.org/x/time v0.15.0 // indirect - golang.org/x/tools v0.43.0 // indirect + golang.org/x/tools v0.44.0 // indirect google.golang.org/api v0.276.0 // indirect google.golang.org/genai v1.54.0 // indirect - google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 // indirect + google.golang.org/genproto/googleapis/rpc v0.0.0-20260427160629-7cedc36a6bc4 // indirect google.golang.org/grpc v1.80.0 // indirect google.golang.org/protobuf v1.36.11 // indirect gopkg.in/dnaeon/go-vcr.v4 v4.0.6-0.20251110073552-01de4eb40290 // indirect gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 // indirect gopkg.in/warnings.v0 v0.1.2 // indirect gopkg.in/yaml.v2 v2.4.0 // indirect - modernc.org/libc v1.72.0 // indirect + modernc.org/libc v1.72.1 // indirect modernc.org/mathutil v1.7.1 // indirect modernc.org/memory v1.11.0 // indirect ) diff --git a/go.sum b/go.sum index 8ce3b215354e3a09b4d951e4769e81d0568d3ee0..3f52470038f05e7a5f8ba7933fccd67117aba6db 100644 --- a/go.sum +++ b/go.sum @@ -2,12 +2,12 @@ charm.land/bubbles/v2 v2.1.0 h1:YSnNh5cPYlYjPxRrzs5VEn3vwhtEn3jVGRBT3M7/I0g= charm.land/bubbles/v2 v2.1.0/go.mod h1:l97h4hym2hvWBVfmJDtrEHHCtkIKeTEb3TTJ4ZOB3wY= charm.land/bubbletea/v2 v2.0.6 h1:UHN/91OyuhaOFGSrBXQ/hMZD8IO1Uc4BvHlgHXL2WJo= charm.land/bubbletea/v2 v2.0.6/go.mod h1:MH/D8ZLlN3op37vQvijKuU29g3rqTp+aQapURFonF9g= -charm.land/catwalk v0.37.14 h1:PYz9777U3vnLgHQWriQA6lHwSSZLBnyvCF77E46W02M= -charm.land/catwalk v0.37.14/go.mod h1:IvMXm7qMKvYuWWlN5A4dA0bTgYg4clG1GAeU9/NlY0w= +charm.land/catwalk v0.39.5 h1:YH64kRKGZyPJeAwTfguO+B8zeltag2n1mPiI1kJTmsM= +charm.land/catwalk v0.39.5/go.mod h1:IvMXm7qMKvYuWWlN5A4dA0bTgYg4clG1GAeU9/NlY0w= charm.land/fang/v2 v2.0.1 h1:zQCM8JQJ1JnQX/66B5jlCYBUxL2as5JXQZ2KJ6EL0mY= charm.land/fang/v2 v2.0.1/go.mod h1:S1GmkpcvK+OB5w9caywUnJcsMew45Ot8FXqoz8ALrII= -charm.land/fantasy v0.21.0 h1:fYeW5axjn7KxJFvXavYhToZDG83zM+or1XEpHqX/GAo= -charm.land/fantasy v0.21.0/go.mod h1:GYYvvDAS3u/Wpb5hX0VxCJPhQCaffHNNeBRtGw04IBI= +charm.land/fantasy v0.23.0 h1:pocjwC5CxfEg1Bpwb0raML2d5ijo3op33Mmd6hYJyo4= +charm.land/fantasy v0.23.0/go.mod h1:4yzSsd9XmFEVjRnF1P0LTEbLTmQX6OLnPkrHaf7iruo= charm.land/glamour/v2 v2.0.0 h1:IDBoqLEy7Hdpb9VOXN+khLP/XSxtJy1VsHuW/yF87+U= charm.land/glamour/v2 v2.0.0/go.mod h1:kjq9WB0s8vuUYZNYey2jp4Lgd9f4cKdzAw88FZtpj/w= charm.land/lipgloss/v2 v2.0.3 h1:yM2zJ4Cf5Y51b7RHIwioil4ApI/aypFXXVHSwlM6RzU= @@ -47,8 +47,8 @@ github.com/PuerkitoBio/goquery v1.12.0 h1:pAcL4g3WRXekcB9AU/y1mbKez2dbY2AajVhtkO github.com/PuerkitoBio/goquery v1.12.0/go.mod h1:802ej+gV2y7bbIhOIoPY5sT183ZW0YFofScC4q/hIpQ= github.com/alecthomas/assert/v2 v2.11.0 h1:2Q9r3ki8+JYXvGsDyBXwH3LcJ+WK5D0gc5E8vS6K3D0= github.com/alecthomas/assert/v2 v2.11.0/go.mod h1:Bze95FyfUr7x34QZrjL+XP+0qgp/zg8yS+TtBj1WA3k= -github.com/alecthomas/chroma/v2 v2.23.1 h1:nv2AVZdTyClGbVQkIzlDm/rnhk1E9bU9nXwmZ/Vk/iY= -github.com/alecthomas/chroma/v2 v2.23.1/go.mod h1:NqVhfBR0lte5Ouh3DcthuUCTUpDC9cxBOfyMbMQPs3o= +github.com/alecthomas/chroma/v2 v2.24.1 h1:m5ffpfZbIb++k8AqFEKy9uVgY12xIQtBsQlc6DfZJQM= +github.com/alecthomas/chroma/v2 v2.24.1/go.mod h1:l+ohZ9xRXIbGe7cIW+YZgOGbvuVLjMps/FYN/CwuabI= github.com/alecthomas/repr v0.5.2 h1:SU73FTI9D1P5UNtvseffFSGmdNci/O6RsqzeXJtP0Qs= github.com/alecthomas/repr v0.5.2/go.mod h1:Fr0507jx4eOXV7AlPV6AVZLYrLIuIeSOWtW57eE/O/4= github.com/andybalholm/cascadia v1.3.2/go.mod h1:7gtRlve5FxPPgIgX36uWBX58OdBsSS6lUvCFb+h7KvU= @@ -58,8 +58,8 @@ github.com/atotto/clipboard v0.1.4 h1:EH0zSVneZPSuFR11BlR9YppQTVDbh5+16AmcJi4g1z github.com/atotto/clipboard v0.1.4/go.mod h1:ZY9tmq7sm5xIbd9bOK4onWV4S6X0u6GY7Vn0Yu86PYI= github.com/aws/aws-sdk-go-v2 v1.41.6 h1:1AX0AthnBQzMx1vbmir3Y4WsnJgiydmnJjiLu+LvXOg= github.com/aws/aws-sdk-go-v2 v1.41.6/go.mod h1:dy0UzBIfwSeot4grGvY1AqFWN5zgziMmWGzysDnHFcQ= -github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.8 h1:eBMB84YGghSocM7PsjmmPffTa+1FBUeNvGvFou6V/4o= -github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.8/go.mod h1:lyw7GFp3qENLh7kwzf7iMzAxDn+NzjXEAGjKS2UOKqI= +github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.9 h1:adBsCIIpLbLmYnkQU+nAChU5yhVTvu5PerROm+/Kq2A= +github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.9/go.mod h1:uOYhgfgThm/ZyAuJGNQ5YgNyOlYfqnGpTHXvk3cpykg= github.com/aws/aws-sdk-go-v2/config v1.32.16 h1:Q0iQ7quUgJP0F/SCRTieScnaMdXr9h/2+wze1u3cNeM= github.com/aws/aws-sdk-go-v2/config v1.32.16/go.mod h1:duCCnJEFqpt2RC6no1iK6q+8HpwOAkiUua0pY507dQc= github.com/aws/aws-sdk-go-v2/credentials v1.19.15 h1:fyvgWTszojq8hEnMi8PPBTvZdTtEVmAVyo+NFLHBhH4= @@ -84,8 +84,8 @@ github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.20 h1:oK/njaL8GtyEihkWMD4k3Vg github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.20/go.mod h1:JHs8/y1f3zY7U5WcuzoJ/yAYGYtNIVPKLIbp61euvmg= github.com/aws/aws-sdk-go-v2/service/sts v1.42.0 h1:ks8KBcZPh3PYISr5dAiXCM5/Thcuxk8l+PG4+A0exds= github.com/aws/aws-sdk-go-v2/service/sts v1.42.0/go.mod h1:pFw33T0WLvXU3rw1WBkpMlkgIn54eCB5FYLhjDc9Foo= -github.com/aws/smithy-go v1.25.0 h1:Sz/XJ64rwuiKtB6j98nDIPyYrV1nVNJ4YU74gttcl5U= -github.com/aws/smithy-go v1.25.0/go.mod h1:YE2RhdIuDbA5E5bTdciG9KrW3+TiEONeUWCqxX9i1Fc= +github.com/aws/smithy-go v1.25.1 h1:J8ERsGSU7d+aCmdQur5Txg6bVoYelvQJgtZehD12GkI= +github.com/aws/smithy-go v1.25.1/go.mod h1:YE2RhdIuDbA5E5bTdciG9KrW3+TiEONeUWCqxX9i1Fc= github.com/aymanbagabas/go-nativeclipboard v0.1.3 h1:FmAWHPTwneAixu7uGDn3cL42xPlUCdNp2J8egMn3P1k= github.com/aymanbagabas/go-nativeclipboard v0.1.3/go.mod h1:2o7MyZwwi4pmXXpOpvOS5FwaHyoCIUks0ktjUvB0EoE= github.com/aymanbagabas/go-udiff v0.4.1 h1:OEIrQ8maEeDBXQDoGCbbTTXYJMYRCRO1fnodZ12Gv5o= @@ -156,8 +156,8 @@ github.com/denisbrodbeck/machineid v1.0.1 h1:geKr9qtkB876mXguW2X6TU4ZynleN6ezuMS github.com/denisbrodbeck/machineid v1.0.1/go.mod h1:dJUwb7PTidGDeYyUBmXZ2GphQBbjJCrnectwCyxcUSI= github.com/disintegration/imaging v1.6.2 h1:w1LecBlG2Lnp8B3jk5zSuNqd7b4DXhcjwek1ei82L+c= github.com/disintegration/imaging v1.6.2/go.mod h1:44/5580QXChDfwIclfc/PCwrr44amcmDAg8hxG0Ewe4= -github.com/dlclark/regexp2 v1.11.5 h1:Q/sSnsKerHeCkc/jSTNq1oCm7KiVgUMZRDUoRu0JQZQ= -github.com/dlclark/regexp2 v1.11.5/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8= +github.com/dlclark/regexp2 v1.12.0 h1:0j4c5qQmnC6XOWNjP3PIXURXN2gWx76rd3KvgdPkCz8= +github.com/dlclark/regexp2 v1.12.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8= github.com/dnaeon/go-vcr v1.2.0 h1:zHCHvJYTMh1N7xnV7zf1m1GPBF9Ad0Jk/whtQ1663qI= github.com/dnaeon/go-vcr v1.2.0/go.mod h1:R4UdLID7HZT3taECzJs4YgbbH6PIGXB6W/sc5OLb6RQ= github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY= @@ -225,8 +225,8 @@ github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6 github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY= github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8= github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU= -github.com/google/jsonschema-go v0.4.2 h1:tmrUohrwoLZZS/P3x7ex0WAVknEkBZM46iALbcqoRA8= -github.com/google/jsonschema-go v0.4.2/go.mod h1:r5quNTdLOYEz95Ru18zA0ydNbBuYoo9tgaYcxEYhJVE= +github.com/google/jsonschema-go v0.4.3 h1:/DBOLZTfDow7pe2GmaJNhltueGTtDKICi8V8p+DQPd0= +github.com/google/jsonschema-go v0.4.3/go.mod h1:r5quNTdLOYEz95Ru18zA0ydNbBuYoo9tgaYcxEYhJVE= github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e h1:ijClszYn+mADRFY17kjQEVQ1XRhq2/JR1M3sGqeJoxs= github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA= github.com/google/s2a-go v0.1.9 h1:LGD7gtMgezd8a/Xak7mEWL0PjoTQFvpRudN895yqKW0= @@ -248,8 +248,8 @@ github.com/hexops/gotextdiff v1.0.3 h1:gitA9+qJrrTCsiCl7+kh75nPqQt1cx4ZkudSTLoUq github.com/hexops/gotextdiff v1.0.3/go.mod h1:pSWU5MAI3yDq+fZBTazCSJysOMbxWL1BSow5/V2vxeg= github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8= github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw= -github.com/invopop/jsonschema v0.13.0 h1:KvpoAJWEjR3uD9Kbm2HWJmqsEaHt8lBUpd0qHcIi21E= -github.com/invopop/jsonschema v0.13.0/go.mod h1:ffZ5Km5SWWRAIN6wbDXItl95euhFz2uON45H2qjYt+0= +github.com/invopop/jsonschema v0.14.0 h1:MHQqLhvpNUZfw+hM3AZDYK7jxO8FZoQeQM77g8iyZjg= +github.com/invopop/jsonschema v0.14.0/go.mod h1:ygm6C2EaVNMBDPpaPlnOA2pFAxBnxGjFlMZABxm9n2I= github.com/itchyny/gojq v0.12.19 h1:ttXA0XCLEMoaLOz5lSeFOZ6u6Q3QxmG46vfgI4O0DEs= github.com/itchyny/gojq v0.12.19/go.mod h1:5galtVPDywX8SPSOrqjGxkBeDhSxEW1gSxoy7tn1iZY= github.com/itchyny/timefmt-go v0.1.8 h1:1YEo1JvfXeAHKdjelbYr/uCuhkybaHCeTkH8Bo791OI= @@ -264,14 +264,14 @@ github.com/jordanella/go-ansi-paintbrush v0.0.0-20240728195301-b7ad996ecf3d h1:o github.com/jordanella/go-ansi-paintbrush v0.0.0-20240728195301-b7ad996ecf3d/go.mod h1:SV0W0APWP9MZ1/gfDQ/NzzTlWdIgYZ/ZbpN4d/UXRYw= github.com/josharian/intern v1.0.0 h1:vlS4z54oSdjm0bgjRigI+G1HpF+tI+9rE5LLzOg8HmY= github.com/josharian/intern v1.0.0/go.mod h1:5DoeVV0s6jJacbCEi61lwdGj/aVlrQvzHFFd8Hwg//Y= -github.com/kaptinlin/go-i18n v0.3.0 h1:wP76dvYg04bvwTb+8NB+CmdZ2kL7lSSCQ9B/kFv7QHo= -github.com/kaptinlin/go-i18n v0.3.0/go.mod h1:pVcu9qsW5pOIOoZFJXesRYmLos1vMQrby70JPAoWmJU= -github.com/kaptinlin/jsonpointer v0.4.17 h1:mY9k8ciWncxbsECyaxKnR0MdmxamNdp2tLQkAKVrtSk= -github.com/kaptinlin/jsonpointer v0.4.17/go.mod h1:SsfsjqnHG5zuKo1DTBzk1VknaHlL4osHw+X9kZKukpU= -github.com/kaptinlin/jsonschema v0.7.7 h1:41BlQJ9dskH0oE5DSzBUrl/w4JQYIr6N6L0B5GNyDoM= -github.com/kaptinlin/jsonschema v0.7.7/go.mod h1:rKjWfyySHSxAD7Li2ctYkPlOu960igoKBvZ2ADRtd5Q= -github.com/kaptinlin/messageformat-go v0.4.19 h1:A5kuuZ1ybXDQ7kD1aoEWGAOemX7hLsMY0yolgSbgpRI= -github.com/kaptinlin/messageformat-go v0.4.19/go.mod h1:utSDTfiXTxl66OC5RIEuObLH7Ue3YjbA2X86SYMBYWg= +github.com/kaptinlin/go-i18n v0.4.4 h1:3XrUYyLOykcd1K3gm4j7ndrF8YLIYrJjtbKGr/nF2Kw= +github.com/kaptinlin/go-i18n v0.4.4/go.mod h1:mU/7BH4molY5lGZYBwBRKAaiJ70dWRHuqmQ0/pFLGno= +github.com/kaptinlin/jsonpointer v0.4.19 h1:dEkwEnvn9jJCofrwKGxfKaPNbDOQEf3UEbEumn4xZBg= +github.com/kaptinlin/jsonpointer v0.4.19/go.mod h1:Mo7+DX8RlQTFqS4dnYJl0izSP4ob+Rl5xO/mGDETgaU= +github.com/kaptinlin/jsonschema v0.7.11 h1:h63Lb3Q4FBSWeWiAGefNPEVPNsOvgn91ATmf25X0yRs= +github.com/kaptinlin/jsonschema v0.7.11/go.mod h1:cJ8QIhwq3V/Yyh3sXRNt8w3sM943bNIbwnPTpBTXn3s= +github.com/kaptinlin/messageformat-go v0.6.0 h1:D6jiXFsKW4/JG2CMddv/F6Rev9KVbCRKEzzV5QOAcpc= +github.com/kaptinlin/messageformat-go v0.6.0/go.mod h1:NKjwS6e9u7DRhAK+vydjDDwJ7UbdHhYjk/yk2WPuZPs= github.com/klauspost/compress v1.18.5 h1:/h1gH5Ce+VWNLSWqPzOVn6XBO+vJbCNGvjoaGBFW2IE= github.com/klauspost/compress v1.18.5/go.mod h1:cwPg85FWrGar70rWktvGQj8/hthj3wpl0PGDogxkrSQ= github.com/klauspost/cpuid/v2 v2.2.10 h1:tBs3QSyvjDyFTq3uoc/9xFpCuOsJQFNPiAhYdw2skhE= @@ -294,8 +294,8 @@ github.com/mailru/easyjson v0.0.0-20190626092158-b2ccc519800e/go.mod h1:C1wdFJiN github.com/mailru/easyjson v0.7.6/go.mod h1:xzfreul335JAWq5oZzymOObrkdz5UnU4kGfJJLY9Nlc= github.com/mailru/easyjson v0.7.7 h1:UGYAvKxe3sBsEDzO8ZeWOSlIQfWFlxbzLZe7hwFURr0= github.com/mailru/easyjson v0.7.7/go.mod h1:xzfreul335JAWq5oZzymOObrkdz5UnU4kGfJJLY9Nlc= -github.com/mattn/go-isatty v0.0.21 h1:xYae+lCNBP7QuW4PUnNG61ffM4hVIfm+zUzDuSzYLGs= -github.com/mattn/go-isatty v0.0.21/go.mod h1:ZXfXG4SQHsB/w3ZeOYbR0PrPwLy+n6xiMrJlRFqopa4= +github.com/mattn/go-isatty v0.0.22 h1:j8l17JJ9i6VGPUFUYoTUKPSgKe/83EYU2zBC7YNKMw4= +github.com/mattn/go-isatty v0.0.22/go.mod h1:ZXfXG4SQHsB/w3ZeOYbR0PrPwLy+n6xiMrJlRFqopa4= github.com/mattn/go-runewidth v0.0.23 h1:7ykA0T0jkPpzSvMS5i9uoNn2Xy3R383f9HDx3RybWcw= github.com/mattn/go-runewidth v0.0.23/go.mod h1:XBkDxAl56ILZc9knddidhrOlY5R/pDhgLpndooCuJAs= github.com/mfridman/interpolate v0.0.2 h1:pnuTK7MQIxxFz1Gr+rjSIx9u7qVjf5VOoM/u6BbAxPY= @@ -304,8 +304,8 @@ github.com/microcosm-cc/bluemonday v1.0.27 h1:MpEUotklkwCSLeH+Qdx1VJgNqLlpY2KXwX github.com/microcosm-cc/bluemonday v1.0.27/go.mod h1:jFi9vgW+H7c3V0lb6nR74Ib/DIB5OBs92Dimizgw2cA= github.com/mitchellh/mapstructure v1.5.0 h1:jeMsZIYE/09sWLaz43PL7Gy6RuMjD2eJVyuac5Z2hdY= github.com/mitchellh/mapstructure v1.5.0/go.mod h1:bFUtVrKA4DC2yAKiSyO/QUcy7e+RRV2QTWOzhPopBRo= -github.com/modelcontextprotocol/go-sdk v1.5.0 h1:CHU0FIX9kpueNkxuYtfYQn1Z0slhFzBZuq+x6IiblIU= -github.com/modelcontextprotocol/go-sdk v1.5.0/go.mod h1:gggDIhoemhWs3BGkGwd1umzEXCEMMvAnhTrnbXJKKKA= +github.com/modelcontextprotocol/go-sdk v1.6.0 h1:PPLS3kn7WtOEnR+Af4X5H96SG0qSab8R/ZQT/HkhPkY= +github.com/modelcontextprotocol/go-sdk v1.6.0/go.mod h1:kzm3kzFL1/+AziGOE0nUs3gvPoNxMCvkxokMkuFapXQ= github.com/muesli/cancelreader v0.2.2 h1:3I4Kt4BQjOR54NavqnDogx/MIoWBFa0StPA8ELUXHmA= github.com/muesli/cancelreader v0.2.2/go.mod h1:3XuTXfFS2VjM+HTLZY9Ak0l6eUKfijIfMUZ4EgX0QYo= github.com/muesli/mango v0.1.0 h1:DZQK45d2gGbql1arsYA4vfg4d7I9Hfx5rX/GCmzsAvI= @@ -316,10 +316,10 @@ github.com/muesli/mango-pflag v0.1.0 h1:UADqbYgpUyRoBja3g6LUL+3LErjpsOwaC9ywvBWe github.com/muesli/mango-pflag v0.1.0/go.mod h1:YEQomTxaCUp8PrbhFh10UfbhbQrM/xJ4i2PB8VTLLW0= github.com/muesli/roff v0.1.0 h1:YD0lalCotmYuF5HhZliKWlIx7IEhiXeSfq7hNjFqGF8= github.com/muesli/roff v0.1.0/go.mod h1:pjAHQM9hdUUwm/krAfrLGgJkXJ+YuhtsfZ42kieB2Ig= -github.com/ncruces/go-sqlite3 v0.33.3 h1:6jCR3KuGvJSEwhaQrkHDGeIe2qCQ6nOUDNsPz7ZIotw= -github.com/ncruces/go-sqlite3 v0.33.3/go.mod h1:t2Osfw0wcKzJTgv2EvrkTtVLqlbKTA5Yvwb2ypAlBcY= -github.com/ncruces/go-sqlite3-wasm v1.1.1-0.20260409221933-87e4b35a38d0 h1:ymE9H30x1AyW5VfMNkJC9teuI2W1jjMsQS7kc6zl6Tg= -github.com/ncruces/go-sqlite3-wasm v1.1.1-0.20260409221933-87e4b35a38d0/go.mod h1:/H3+JykPsfSlvKbOxNSx9kKwm3ecqQGzyCs1e9KkNsU= +github.com/ncruces/go-sqlite3 v0.34.0 h1:q2I6wHTLWIoz6ehYkKdG5dGQc66eJv7ZGnekhvuMfK8= +github.com/ncruces/go-sqlite3 v0.34.0/go.mod h1:qpBxsSdGPnO9K5OExuv5GEsrGQ7Rk6JsJFH6wn2DwwU= +github.com/ncruces/go-sqlite3-wasm/v2 v2.1.35300 h1:cRdxCt3BDfMu0vfSdoqaAPD+dzIXPkGREjqyZMLN2Ak= +github.com/ncruces/go-sqlite3-wasm/v2 v2.1.35300/go.mod h1:R2kJLPoSA/GBX/b8x7zwOq/KLAw6rLMY1l3Hi76SQIo= github.com/ncruces/go-strftime v1.0.0 h1:HMFp8mLCTPp341M/ZnA4qaf7ZlsbTc+miZjCLOFAw7w= github.com/ncruces/go-strftime v1.0.0/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls= github.com/ncruces/julianday v1.0.0 h1:fH0OKwa7NWvniGQtxdJRxAgkBMolni2BjDHaWTxqt7M= @@ -331,10 +331,12 @@ github.com/nxadm/tail v1.4.11 h1:8feyoE3OzPrcshW5/MJ4sGESc5cqmGkGCWlco4l0bqY= github.com/nxadm/tail v1.4.11/go.mod h1:OTaG3NK980DZzxbRq6lEuzgU+mug70nY11sMd4JXXHc= github.com/onsi/gomega v1.34.1 h1:EUMJIKUjM8sKjYbtxQI9A4z2o+rruxnzNvpknOXie6k= github.com/onsi/gomega v1.34.1/go.mod h1:kU1QgUvBDLXBJq618Xvm2LUX6rSAfRaFRTcdOeDLwwY= -github.com/pelletier/go-toml/v2 v2.2.4 h1:mye9XuhQ6gvn5h28+VilKrrPoQVanw5PMw/TB0t5Ec4= -github.com/pelletier/go-toml/v2 v2.2.4/go.mod h1:2gIqNv+qfxSVS7cM2xJQKtLSTLUE9V8t9Stt+h56mCY= -github.com/pierrec/lz4/v4 v4.1.25 h1:kocOqRffaIbU5djlIBr7Wh+cx82C0vtFb0fOurZHqD0= -github.com/pierrec/lz4/v4 v4.1.25/go.mod h1:EoQMVJgeeEOMsCqCzqFm2O0cJvljX2nGZjcRIPL34O4= +github.com/pb33f/ordered-map/v2 v2.3.1 h1:5319HDO0aw4DA4gzi+zv4FXU9UlSs3xGZ40wcP1nBjY= +github.com/pb33f/ordered-map/v2 v2.3.1/go.mod h1:qxFQgd0PkVUtOMCkTapqotNgzRhMPL7VvaHKbd1HnmQ= +github.com/pelletier/go-toml/v2 v2.3.0 h1:k59bC/lIZREW0/iVaQR8nDHxVq8OVlIzYCOJf421CaM= +github.com/pelletier/go-toml/v2 v2.3.0/go.mod h1:2gIqNv+qfxSVS7cM2xJQKtLSTLUE9V8t9Stt+h56mCY= +github.com/pierrec/lz4/v4 v4.1.26 h1:GrpZw1gZttORinvzBdXPUXATeqlJjqUG/D87TKMnhjY= +github.com/pierrec/lz4/v4 v4.1.26/go.mod h1:EoQMVJgeeEOMsCqCzqFm2O0cJvljX2nGZjcRIPL34O4= github.com/pjbgf/sha1cd v0.3.2 h1:a9wb0bp1oC2TGwStyn0Umc/IGKQnEgF0vVaZ8QF8eo4= github.com/pjbgf/sha1cd v0.3.2/go.mod h1:zQWigSxVmsHEZow5qaLtPYxpcKMMQpa09ixqBxuCS6A= github.com/pkg/browser v0.0.0-20240102092130-5ac0b6a4141c h1:+mdjkGKdHQG3305AYmdv1U2eRNDiU2ErMBj1gwrq8eQ= @@ -347,10 +349,10 @@ github.com/planetscale/vtprotobuf v0.6.1-0.20240319094008-0393e58bdf10/go.mod h1 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4= github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U= github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4= -github.com/posthog/posthog-go v1.12.1 h1:qZMHfC0frQOR1LT4js3ns+pXbDIyFsV+kWpvJEok3ms= -github.com/posthog/posthog-go v1.12.1/go.mod h1:xsVOW9YImilUcazwPNEq4PJDqEZf2KeCS758zXjwkPg= -github.com/pressly/goose/v3 v3.27.0 h1:/D30gVTuQhu0WsNZYbJi4DMOsx1lNq+6SkLe+Wp59BM= -github.com/pressly/goose/v3 v3.27.0/go.mod h1:3ZBeCXqzkgIRvrEMDkYh1guvtoJTU5oMMuDdkutoM78= +github.com/posthog/posthog-go v1.12.4 h1:fAqTAGgLQFZADRg3AJvSSoP0hwEIs940OE8AmL8/H8c= +github.com/posthog/posthog-go v1.12.4/go.mod h1:xsVOW9YImilUcazwPNEq4PJDqEZf2KeCS758zXjwkPg= +github.com/pressly/goose/v3 v3.27.1 h1:6uEvcprBybDmW4hcz3gYujhARhye+GoWKhEWyzD5sh4= +github.com/pressly/goose/v3 v3.27.1/go.mod h1:maruOxsPnIG2yHHyo8UqKWXYKFcH7Q76csUV7+7KYoM= github.com/qjebbs/go-jsons v1.0.0-alpha.5 h1:U2PPDxeKI1MMOSw7e7xyxhwH9Ggc7UrDvaRIkJ+l0n8= github.com/qjebbs/go-jsons v1.0.0-alpha.5/go.mod h1:wNJrtinHyC3YSf6giEh4FJN8+yZV7nXBjvmfjhBIcw4= github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE= @@ -417,8 +419,6 @@ github.com/u-root/u-root v0.14.1-0.20250807200646-5e7721023dc7 h1:ax+jBy7xFhh+Ka github.com/u-root/u-root v0.14.1-0.20250807200646-5e7721023dc7/go.mod h1:/0Qr7qJeDwWxoKku2xKQ4Szc+SwBE3g9VE8jNiamsmc= github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701 h1:pyC9PaHYZFgEKFdlp3G8RaCKgVpHZnecvArXvPXcFkM= github.com/u-root/uio v0.0.0-20240224005618-d2acac8f3701/go.mod h1:P3a5rG4X7tI17Nn3aOIAYr5HbIMukwXG0urG0WuL8OA= -github.com/wk8/go-ordered-map/v2 v2.1.8 h1:5h/BUHu93oj4gIdvHHHGsScSTMijfx5PeYkE/fJgbpc= -github.com/wk8/go-ordered-map/v2 v2.1.8/go.mod h1:5nJHM5DyteebpVlHnWMV0rPz6Zp7+xBAnxjb1X5vnTw= github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e h1:JVG44RsyaB9T2KIHavMF/ppJZNG9ZpyihvCd0w101no= github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e/go.mod h1:RbqR21r5mrJuqunuUZ/Dhy/avygyECGrLceyNeo4LiM= github.com/yosida95/uritemplate/v3 v3.0.2 h1:Ed3Oyj9yrmi9087+NczuL5BwkIc4wvTb5zIM+UJPGz4= @@ -465,8 +465,8 @@ golang.org/x/crypto v0.23.0/go.mod h1:CKFgDieR+mRhux2Lsu27y0fO304Db0wZe70UKqHu0v golang.org/x/crypto v0.31.0/go.mod h1:kDsLvtWBEx7MV9tJOj9bnXsPbxwJQ6csT/x4KIN4Ssk= golang.org/x/crypto v0.50.0 h1:zO47/JPrL6vsNkINmLoo/PH1gcxpls50DNogFvB5ZGI= golang.org/x/crypto v0.50.0/go.mod h1:3muZ7vA7PBCE6xgPX7nkzzjiUq87kRItoJQM1Yo8S+Q= -golang.org/x/exp v0.0.0-20260312153236-7ab1446f8b90 h1:jiDhWWeC7jfWqR9c/uplMOqJ0sbNlNWv0UkzE0vX1MA= -golang.org/x/exp v0.0.0-20260312153236-7ab1446f8b90/go.mod h1:xE1HEv6b+1SCZ5/uscMRjUBKtIxworgEcEi+/n9NQDQ= +golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f h1:W3F4c+6OLc6H2lb//N1q4WpJkhzJCK5J6kUi1NTVXfM= +golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f/go.mod h1:J1xhfL/vlindoeF/aINzNzt2Bket5bjo9sdOYzOsU80= golang.org/x/image v0.0.0-20191009234506-e7c1f5e7dbb8/go.mod h1:FeLwcggjj3mMvU+oOTbSwawSJRM1uh48EjtB4UJZlP0= golang.org/x/image v0.38.0 h1:5l+q+Y9JDC7mBOMjo4/aPhMDcxEptsX+Tt3GgRQRPuE= golang.org/x/image v0.38.0/go.mod h1:/3f6vaXC+6CEanU4KJxbcUZyEePbyKbaLoDOe4ehFYY= @@ -475,8 +475,8 @@ golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs= golang.org/x/mod v0.12.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs= golang.org/x/mod v0.15.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c= golang.org/x/mod v0.17.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c= -golang.org/x/mod v0.34.0 h1:xIHgNUUnW6sYkcM5Jleh05DvLOtwc6RitGHbDk4akRI= -golang.org/x/mod v0.34.0/go.mod h1:ykgH52iCZe79kzLLMhyCUzhMci+nQj+0XkbXpNYtVjY= +golang.org/x/mod v0.35.0 h1:Ww1D637e6Pg+Zb2KrWfHQUnH2dQRLBQyAtpr/haaJeM= +golang.org/x/mod v0.35.0/go.mod h1:+GwiRhIInF8wPm+4AoT6L0FA1QWAad3OMdTRx4tFYlU= golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s= golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg= golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c= @@ -550,8 +550,8 @@ golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU= golang.org/x/tools v0.13.0/go.mod h1:HvlwmtVNQAhOuCjW7xxvovg8wbNq7LwfXh/k7wXUl58= golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d/go.mod h1:aiJjzUbINMkxbQROHiO6hDPo2LHcIPhhQsa9DLh0yGk= -golang.org/x/tools v0.43.0 h1:12BdW9CeB3Z+J/I/wj34VMl8X+fEXBxVR90JeMX5E7s= -golang.org/x/tools v0.43.0/go.mod h1:uHkMso649BX2cZK6+RpuIPXS3ho2hZo4FVwfoy1vIk0= +golang.org/x/tools v0.44.0 h1:UP4ajHPIcuMjT1GqzDWRlalUEoY+uzoZKnhOjbIPD2c= +golang.org/x/tools v0.44.0/go.mod h1:KA0AfVErSdxRZIsOVipbv3rQhVXTnlU6UhKxHd1seDI= golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4= gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E= @@ -559,12 +559,12 @@ google.golang.org/api v0.276.0 h1:nVArUtfLEihtW+b0DdcqRGK1xoEm2+ltAihyztq7MKY= google.golang.org/api v0.276.0/go.mod h1:Fnag/EWUPIcJXuIkP1pjoTgS5vdxlk3eeemL7Do6bvw= google.golang.org/genai v1.54.0 h1:ZQCa70WMTJDI11FdqWCzGvZ5PanpcpfoO6jl/lrSnGU= google.golang.org/genai v1.54.0/go.mod h1:A3kkl0nyBjyFlNjgxIwKq70julKbIxpSxqKO5gw/gmk= -google.golang.org/genproto v0.0.0-20260414002931-afd174a4e478 h1:aLsVTW0lZ8+IY5u/ERjZSCvAmhuR7slKzyha3YikDNA= -google.golang.org/genproto v0.0.0-20260414002931-afd174a4e478/go.mod h1:YJAzKjfHIUHb9T+bfu8L7mthAp7VVXQBUs1PLdBWS7M= -google.golang.org/genproto/googleapis/api v0.0.0-20260414002931-afd174a4e478 h1:yQugLulqltosq0B/f8l4w9VryjV+N/5gcW0jQ3N8Qec= -google.golang.org/genproto/googleapis/api v0.0.0-20260414002931-afd174a4e478/go.mod h1:C6ADNqOxbgdUUeRTU+LCHDPB9ttAMCTff6auwCVa4uc= -google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 h1:RmoJA1ujG+/lRGNfUnOMfhCy5EipVMyvUE+KNbPbTlw= -google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8= +google.golang.org/genproto v0.0.0-20260427160629-7cedc36a6bc4 h1:2iMJZntwvmfgtse+s744JY7v7PgEdSBuFYXucvpOHNM= +google.golang.org/genproto v0.0.0-20260427160629-7cedc36a6bc4/go.mod h1:v14kaaboYyXQ1Gsu489Q+Hg/oN4B33mWtuOhF1HCeXA= +google.golang.org/genproto/googleapis/api v0.0.0-20260427160629-7cedc36a6bc4 h1:yOzSCGPx+cp5VO7IxvZ9SBFF7j1tZVcNtlHR2iYKtVo= +google.golang.org/genproto/googleapis/api v0.0.0-20260427160629-7cedc36a6bc4/go.mod h1:Q9HWtNeE7tM9npdIsEvqXj1QJIvVoeAV3rtXtS715Cw= +google.golang.org/genproto/googleapis/rpc v0.0.0-20260427160629-7cedc36a6bc4 h1:tEkOQcXgF6dH1G+MVKZrfpYvozGrzb91k6ha7jireSM= +google.golang.org/genproto/googleapis/rpc v0.0.0-20260427160629-7cedc36a6bc4/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8= google.golang.org/grpc v1.80.0 h1:Xr6m2WmWZLETvUNvIUmeD5OAagMw3FiKmMlTdViWsHM= google.golang.org/grpc v1.80.0/go.mod h1:ho/dLnxwi3EDJA4Zghp7k2Ec1+c2jqup0bFkw07bwF4= google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE= @@ -590,10 +590,10 @@ gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C gopkg.in/yaml.v3 v3.0.0-20200615113413-eeeca48fe776/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM= gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA= gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM= -modernc.org/cc/v4 v4.27.3 h1:uNCgn37E5U09mTv1XgskEVUJ8ADKpmFMPxzGJ0TSo+U= -modernc.org/cc/v4 v4.27.3/go.mod h1:3YjcbCqhoTTHPycJDRl2WZKKFj0nwcOIPBfEZK0Hdk8= -modernc.org/ccgo/v4 v4.32.4 h1:L5OB8rpEX4ZsXEQwGozRfJyJSFHbbNVOoQ59DU9/KuU= -modernc.org/ccgo/v4 v4.32.4/go.mod h1:lY7f+fiTDHfcv6YlRgSkxYfhs+UvOEEzj49jAn2TOx0= +modernc.org/cc/v4 v4.28.1 h1:XpLbkYVQ24E8tX5u8+yWGvaxerxkR/S4zqxI8ZoSBuc= +modernc.org/cc/v4 v4.28.1/go.mod h1:OnovgIhbbMXMu1aISnJ0wvVD1KnW+cAUJkIrAWh+kVI= +modernc.org/ccgo/v4 v4.33.0 h1:dspBCm75jsj8Y/ufwAMVfe375L2iYdMyQ2QG/v3hL54= +modernc.org/ccgo/v4 v4.33.0/go.mod h1:+RhXBoRYzRwaH21mV/aj6XvQRDtfjcZfAlPMsQo8CR0= modernc.org/fileutil v1.4.0 h1:j6ZzNTftVS054gi281TyLjHPp6CPHr2KCxEXjEbD6SM= modernc.org/fileutil v1.4.0/go.mod h1:EqdKFDxiByqxLk8ozOxObDSfcVOv/54xDs/DUHdvCUU= modernc.org/gc/v2 v2.6.5 h1:nyqdV8q46KvTpZlsw66kWqwXRHdjIlJOhG6kxiV/9xI= @@ -602,18 +602,18 @@ modernc.org/gc/v3 v3.1.2 h1:ZtDCnhonXSZexk/AYsegNRV1lJGgaNZJuKjJSWKyEqo= modernc.org/gc/v3 v3.1.2/go.mod h1:HFK/6AGESC7Ex+EZJhJ2Gni6cTaYpSMmU/cT9RmlfYY= modernc.org/goabi0 v0.2.0 h1:HvEowk7LxcPd0eq6mVOAEMai46V+i7Jrj13t4AzuNks= modernc.org/goabi0 v0.2.0/go.mod h1:CEFRnnJhKvWT1c1JTI3Avm+tgOWbkOu5oPA8eH8LnMI= -modernc.org/libc v1.72.0 h1:IEu559v9a0XWjw0DPoVKtXpO2qt5NVLAnFaBbjq+n8c= -modernc.org/libc v1.72.0/go.mod h1:tTU8DL8A+XLVkEY3x5E/tO7s2Q/q42EtnNWda/L5QhQ= +modernc.org/libc v1.72.1 h1:db1xwJ6u1kE3KHTFTTbe2GCrczHPKzlURP0aDC4NGD0= +modernc.org/libc v1.72.1/go.mod h1:HRMiC/PhPGLIPM7GzAFCbI+oSgE3dhZ8FWftmRrHVlY= modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU= modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg= modernc.org/memory v1.11.0 h1:o4QC8aMQzmcwCK3t3Ux/ZHmwFPzE6hf2Y5LbkRs+hbI= modernc.org/memory v1.11.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw= -modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8= -modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns= +modernc.org/opt v0.2.0 h1:tGyef5ApycA7FSEOMraay9SaTk5zmbx7Tu+cJs4QKZg= +modernc.org/opt v0.2.0/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns= modernc.org/sortutil v1.2.1 h1:+xyoGf15mM3NMlPDnFqrteY07klSFxLElE2PVuWIJ7w= modernc.org/sortutil v1.2.1/go.mod h1:7ZI3a3REbai7gzCLcotuw9AC4VZVpYMjDzETGsSMqJE= -modernc.org/sqlite v1.49.1 h1:dYGHTKcX1sJ+EQDnUzvz4TJ5GbuvhNJa8Fg6ElGx73U= -modernc.org/sqlite v1.49.1/go.mod h1:m0w8xhwYUVY3H6pSDwc3gkJ/irZT/0YEXwBlhaxQEew= +modernc.org/sqlite v1.50.0 h1:eMowQSWLK0MeiQTdmz3lqoF5dqclujdlIKeJA11+7oM= +modernc.org/sqlite v1.50.0/go.mod h1:m0w8xhwYUVY3H6pSDwc3gkJ/irZT/0YEXwBlhaxQEew= modernc.org/strutil v1.2.1 h1:UneZBkQA+DX2Rp35KcM69cSsNES9ly8mQWD71HKlOA0= modernc.org/strutil v1.2.1/go.mod h1:EHkiggD70koQxjVdSBM3JKM7k6L0FbGE5eymy9i3B9A= modernc.org/token v1.1.0 h1:Xl7Ap9dKaEs5kLoOQeQmPWevfnk/DM5qcLcYlA8ys6Y= diff --git a/internal/agent/coordinator.go b/internal/agent/coordinator.go index 6c5afbc418dbf1ec611080a73fe8316400916b01..0050299b8bef46d95a2ec4a5913043917f7b056f 100644 --- a/internal/agent/coordinator.go +++ b/internal/agent/coordinator.go @@ -281,20 +281,7 @@ func getProviderOptions(model Model, providerCfg config.ProviderConfig) fantasy. return options } - providerType := providerCfg.Type - if providerType == "hyper" { - if strings.Contains(model.CatwalkCfg.ID, "claude") { - providerType = anthropic.Name - } else if strings.Contains(model.CatwalkCfg.ID, "gpt") { - providerType = openai.Name - } else if strings.Contains(model.CatwalkCfg.ID, "gemini") { - providerType = google.Name - } else { - providerType = openaicompat.Name - } - } - - switch providerType { + switch providerCfg.Type { case openai.Name, azure.Name: _, hasReasoningEffort := mergedOptions["reasoning_effort"] if !hasReasoningEffort && model.ModelCfg.ReasoningEffort != "" { @@ -374,11 +361,39 @@ func getProviderOptions(model Model, providerCfg config.ProviderConfig) fantasy. if err == nil { options[google.Name] = parsed } - case openaicompat.Name: + case openaicompat.Name, hyper.Name: _, hasReasoningEffort := mergedOptions["reasoning_effort"] if !hasReasoningEffort && model.ModelCfg.ReasoningEffort != "" { mergedOptions["reasoning_effort"] = model.ModelCfg.ReasoningEffort } + + extraBody := make(map[string]any) + + // "reasoning effort" is a standard OpenAI field, but "thinking" is not. + // Setting it in the right way for each provider. + // TODO: Abstract this in Fantasy somehow? + // TODO: Allow custom providers to specify how to set this? + switch providerCfg.ID { + case hyper.Name: + extraBody["thinking"] = model.ModelCfg.Think + case string(catwalk.InferenceProviderIoNet): + extraBody["chat_template_kwargs"] = map[string]any{ + "thinking": model.ModelCfg.Think, + } + case string(catwalk.InferenceProviderZAI): + if model.ModelCfg.Think { + extraBody["thinking"] = map[string]any{ + "type": "enabled", + } + } else { + extraBody["thinking"] = map[string]any{ + "type": "disabled", + } + } + } + + mergedOptions["extra_body"] = extraBody + parsed, err := openaicompat.ParseOptions(mergedOptions) if err == nil { options[openaicompat.Name] = parsed diff --git a/internal/agent/hyper/provider.go b/internal/agent/hyper/provider.go index d6be84a3dcaa6c08b30ce23f945458c03637a4b4..c8acf132db59de4cbd8c70936c41085cf3758793 100644 --- a/internal/agent/hyper/provider.go +++ b/internal/agent/hyper/provider.go @@ -3,34 +3,24 @@ package hyper import ( "cmp" + "context" _ "embed" "encoding/json" + "fmt" "log/slog" + "net/http" "os" - "strconv" "sync" + "time" "charm.land/catwalk/pkg/catwalk" ) -//go:generate wget -O provider.json https://hyper.charm.land/api/v1/provider +//go:generate wget -O provider.json https://hyper.charm.land/v1/provider //go:embed provider.json var embedded []byte -// Enabled returns true if hyper is enabled. -var Enabled = sync.OnceValue(func() bool { - b, _ := strconv.ParseBool( - cmp.Or( - os.Getenv("HYPER"), - os.Getenv("HYPERCRUSH"), - os.Getenv("HYPER_ENABLE"), - os.Getenv("HYPER_ENABLED"), - ), - ) - return b -}) - // Embedded returns the embedded Hyper provider. var Embedded = sync.OnceValue(func() catwalk.Provider { var provider catwalk.Provider @@ -46,6 +36,8 @@ var Embedded = sync.OnceValue(func() catwalk.Provider { const ( // Name is the default name of this meta provider. Name = "hyper" + // DisplayName is the display name of Hyper. + DisplayName = "Charm Hyper" // defaultBaseURL is the default proxy URL. defaultBaseURL = "https://hyper.charm.land" ) @@ -54,3 +46,38 @@ const ( var BaseURL = sync.OnceValue(func() string { return cmp.Or(os.Getenv("HYPER_URL"), defaultBaseURL) }) + +// FetchCredits calls the Hyper /v1/credits endpoint and returns the remaining +// credits count. +func FetchCredits(ctx context.Context, apiKey string) (int, error) { + req, err := http.NewRequestWithContext( + ctx, + http.MethodGet, + BaseURL()+"/v1/credits", + nil, + ) + if err != nil { + return 0, fmt.Errorf("could not create request: %w", err) + } + req.Header.Set("Authorization", "Bearer "+apiKey) + + client := &http.Client{Timeout: 10 * time.Second} + resp, err := client.Do(req) + if err != nil { + return 0, fmt.Errorf("failed to make request: %w", err) + } + defer resp.Body.Close() //nolint:errcheck + + if resp.StatusCode != http.StatusOK { + return 0, fmt.Errorf("unexpected status code: %d", resp.StatusCode) + } + + var result struct { + Balance int `json:"balance"` + } + if err := json.NewDecoder(resp.Body).Decode(&result); err != nil { + return 0, fmt.Errorf("failed to decode response: %w", err) + } + + return result.Balance, nil +} diff --git a/internal/agent/hyper/provider.json b/internal/agent/hyper/provider.json index 031b03c89ff15b33f252db0aac75591e65adce5b..9bea686175a8135c8b6baf8b1adc0213c2535dd5 100644 --- a/internal/agent/hyper/provider.json +++ b/internal/agent/hyper/provider.json @@ -6,6 +6,42 @@ "default_large_model_id": "kimi-k2.5", "default_small_model_id": "gpt-oss-120b", "models": [ + { + "id": "deepseek-v4-flash", + "name": "DeepSeek V4 Flash", + "cost_per_1m_in": 1.55, + "cost_per_1m_out": 2.28, + "cost_per_1m_in_cached": 0.38, + "cost_per_1m_out_cached": 0, + "context_window": 65000, + "default_max_tokens": 6500, + "can_reason": true, + "supports_attachments": false + }, + { + "id": "deepseek-v4-pro", + "name": "DeepSeek V4 Pro", + "cost_per_1m_in": 4.45, + "cost_per_1m_out": 5.5, + "cost_per_1m_in_cached": 0.35, + "cost_per_1m_out_cached": 0, + "context_window": 600000, + "default_max_tokens": 60000, + "can_reason": true, + "supports_attachments": false + }, + { + "id": "gemma-4-26b-a4b-it", + "name": "Gemma 4 26B A4B", + "cost_per_1m_in": 0.145, + "cost_per_1m_out": 0.5, + "cost_per_1m_in_cached": 0.08, + "cost_per_1m_out_cached": 0, + "context_window": 262144, + "default_max_tokens": 26214, + "can_reason": true, + "supports_attachments": false + }, { "id": "glm-5", "name": "GLM-5", @@ -21,7 +57,7 @@ { "id": "glm-5.1", "name": "GLM-5.1", - "cost_per_1m_in": 1.06, + "cost_per_1m_in": 1.5, "cost_per_1m_out": 4.4, "cost_per_1m_in_cached": 0.26, "cost_per_1m_out_cached": 0, @@ -63,14 +99,74 @@ { "id": "kimi-k2.6", "name": "Kimi K2.6", - "cost_per_1m_in": 0.8, - "cost_per_1m_out": 4, - "cost_per_1m_in_cached": 0.25, + "cost_per_1m_in": 0.74, + "cost_per_1m_out": 3.49, + "cost_per_1m_in_cached": 0.14, "cost_per_1m_out_cached": 0, - "context_window": 262144, + "context_window": 262142, "default_max_tokens": 26214, "can_reason": true, "supports_attachments": true + }, + { + "id": "llama-3.3-70b-instruct", + "name": "Llama 3.3 70B Instruct", + "cost_per_1m_in": 0.1, + "cost_per_1m_out": 0.32, + "cost_per_1m_in_cached": 0.05, + "cost_per_1m_out_cached": 0.2, + "context_window": 128000, + "default_max_tokens": 12800, + "can_reason": true, + "supports_attachments": false + }, + { + "id": "llama-4-maverick-17b-128e-instruct-fp8", + "name": "Llama 4 Maverick 17B 128E Instruct FP8", + "cost_per_1m_in": 0.15, + "cost_per_1m_out": 0.6, + "cost_per_1m_in_cached": 0.075, + "cost_per_1m_out_cached": 0.3, + "context_window": 430000, + "default_max_tokens": 43000, + "can_reason": true, + "supports_attachments": true + }, + { + "id": "mistral-large-instruct-2411", + "name": "Mistral Large Instruct 2411", + "cost_per_1m_in": 2, + "cost_per_1m_out": 6, + "cost_per_1m_in_cached": 1, + "cost_per_1m_out_cached": 4, + "context_window": 128000, + "default_max_tokens": 12800, + "can_reason": false, + "supports_attachments": true + }, + { + "id": "qwen3-coder-480b-a35b-instruct-int4-mixed-ar", + "name": "Qwen3 Coder 480B A35B Instruct INT4 Mixed AR", + "cost_per_1m_in": 0.22, + "cost_per_1m_out": 0.95, + "cost_per_1m_in_cached": 0.11, + "cost_per_1m_out_cached": 0.44, + "context_window": 106000, + "default_max_tokens": 10600, + "can_reason": false, + "supports_attachments": false + }, + { + "id": "qwen3-next-80b-a3b-instruct", + "name": "Qwen3 Next 80B A3B Instruct", + "cost_per_1m_in": 0.06, + "cost_per_1m_out": 0.6, + "cost_per_1m_in_cached": 0.03, + "cost_per_1m_out_cached": 0.12, + "context_window": 262144, + "default_max_tokens": 26214, + "can_reason": false, + "supports_attachments": false } ] } \ No newline at end of file diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/bash_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/bash_tool.yaml index 9d36f2c102d7a89d85e8481958f19a2e0c03ab7e..45a7a0fe082dec9695ba42693fe54c4006cf4f3e 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/bash_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/bash_tool.yaml @@ -2,6 +2,104 @@ version: 2 interactions: - id: 0 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 33682 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use bash to create a file named test.txt with content 'hello bash'. do not print its timestamp\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" using"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" without"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" printing"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_1128d2594e71440e87395176","type":"function","index":0,"function":{"name":"bash","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"command\":\"echo 'hello bash' \u003e test.txt\",\"description\":\"Create test.txt with content\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-e88f6a52ff032a473793901c","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":709,"completion_tokens":37,"total_tokens":7786}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.665328167s +- id: 1 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 34072 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use bash to create a file named test.txt with content 'hello bash'. do not print its timestamp\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_1128d2594e71440e87395176\",\"function\":{\"arguments\":\"{\\\"command\\\":\\\"echo 'hello bash' > test.txt\\\",\\\"description\\\":\\\"Create test.txt with content\\\"}\",\"name\":\"bash\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Create a file using bash without printing timestamp.\"},{\"content\":\"no output\",\"tool_call_id\":\"call_1128d2594e71440e87395176\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-8545d39e9f867f4ae105f086","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8545d39e9f867f4ae105f086","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Done"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8545d39e9f867f4ae105f086","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-8545d39e9f867f4ae105f086","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":39,"completion_tokens":4,"total_tokens":7787}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.005432041s +- id: 2 request: proto: HTTP/1.1 proto_major: 1 @@ -25,461 +123,669 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" request"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Language"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" English"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" named"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" command"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" do"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" not"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" its"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" must"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"B"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Less"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" than"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" command"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Create"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"B"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" command"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" without"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" without"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" command"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" C"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" x"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" w"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" b"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"47"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" w"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"47"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" okay"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" period"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" u"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"B"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"ash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" command"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} - data: {"id":"chatcmpl-e969f57068607f238298e1c4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":11,"completion_tokens":179,"total_tokens":398}} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 422.115958ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33682 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use bash to create a file named test.txt with content 'hello bash'. do not print its timestamp\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Simple"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" task"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" -"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"46"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"47"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"48"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" specific"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" p"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" using"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" without"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" printing"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" okay"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_83c8d2ffea684ef49a39b7f6","type":"function","index":0,"function":{"name":"bash","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"command\":\"echo 'hello bash' \u003e /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool/test.txt\",\"description\":\"Create test.txt with content\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Could"},"finish_reason":null}]} - data: {"id":"chatcmpl-5ef96ea9d6e37c4f5390151e","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":5,"completion_tokens":60,"total_tokens":7809}} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 8.226233625s -- id: 2 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 34158 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use bash to create a file named test.txt with content 'hello bash'. do not print its timestamp\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_83c8d2ffea684ef49a39b7f6\",\"function\":{\"arguments\":\"{\\\"command\\\":\\\"echo 'hello bash' > /tmp/crush-test/TestCoderAgent/glm-5.1/bash_tool/test.txt\\\",\\\"description\\\":\\\"Create test.txt with content\\\"}\",\"name\":\"bash\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Simple task - create a file with specific content using bash, without printing timestamp.\"},{\"content\":\"no output\",\"tool_call_id\":\"call_83c8d2ffea684ef49a39b7f6\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-d53d988f4caa6753ac1dd6f3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" bash"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" timestamp"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" C"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" x"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" w"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" b"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" h"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ,"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" p"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Create"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" test"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" bash"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-d53d988f4caa6753ac1dd6f3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Done"},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-d53d988f4caa6753ac1dd6f3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" timestamp"},"finish_reason":null}]} - data: {"id":"chatcmpl-d53d988f4caa6753ac1dd6f3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":55,"completion_tokens":4,"total_tokens":7803}} + data: {"id":"chatcmpl-cfd9dbe8bea7d9b3c0cf1225","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":43,"completion_tokens":339,"total_tokens":558}} data: [DONE] @@ -488,4 +794,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 3.896108458s + duration: 591.878125ms diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/download_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/download_tool.yaml index bf6bd114055d530bc6e08f1f816ebf7b0b803cdc..0fdb07b761ee88f3d1ee7c749388cf54f38e4465 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/download_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/download_tool.yaml @@ -25,213 +25,391 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" asks"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" message"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" The"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" download"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" message"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"download"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" https"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"://"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"-files"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" https"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".online"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"://"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"-con"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"example"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"vert"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"-files"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".com"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".online"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/document"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"-con"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/t"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"vert"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"xt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".com"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/document"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/t"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"xt"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/example"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" as"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" as"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" English"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"),"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Download"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Download"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Save"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Download"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Download"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Save"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" D"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fine"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" w"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Return"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" only"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" n"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Download"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Save"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" d"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-d1a5ef493c193e479231adc3","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":14,"completion_tokens":111,"total_tokens":333}} + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" x"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" p"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" x"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" f"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" m"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" U"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" R"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" L"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" d"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" v"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" okay"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fine"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Download"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" example"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" URL"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" save"},"finish_reason":null}]} + + data: {"id":"chatcmpl-8efc292f6882f3c242d54c7e","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":46,"completion_tokens":200,"total_tokens":422}} data: [DONE] @@ -240,7 +418,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 873.598583ms + duration: 747.881583ms - id: 1 request: proto: HTTP/1.1 @@ -248,7 +426,7 @@ interactions: proto_minor: 1 content_length: 33707 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/download_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"download the file from https://example-files.online-convert.com/document/txt/example.txt and save it as example.txt\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/download_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"download the file from https://example-files.online-convert.com/document/txt/example.txt and save it as example.txt\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -265,49 +443,47 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" download"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" download"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" save"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" as"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" as"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_3eceec4c44ad4007af956277","type":"function","index":0,"function":{"name":"download","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_506e856f9cbd4a19a541d177","type":"function","index":0,"function":{"name":"download","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/download_tool/example.txt\",\"url\":\"https://example-files.online-convert.com/document/txt/example.txt\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/download_tool/example.txt\",\"url\":\"https://example-files.online-convert.com/document/txt/example.txt\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-c85e11ddb3887f7f8f5d07a3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":8,"completion_tokens":64,"total_tokens":7816}} + data: {"id":"chatcmpl-e952eac50d9161cbe8a2b906","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":712,"completion_tokens":63,"total_tokens":7815}} data: [DONE] @@ -316,7 +492,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 7.807126583s + duration: 5.040665625s - id: 2 request: proto: HTTP/1.1 @@ -362,15 +538,15 @@ interactions: - text/plain; charset=UTF-8 status: 200 OK code: 200 - duration: 530.746583ms + duration: 1.5467765s - id: 3 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34273 + content_length: 34270 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/download_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"download the file from https://example-files.online-convert.com/document/txt/example.txt and save it as example.txt\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_3eceec4c44ad4007af956277\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/download_tool/example.txt\\\",\\\"url\\\":\\\"https://example-files.online-convert.com/document/txt/example.txt\\\"}\",\"name\":\"download\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to download a file from a URL and save it as example.txt.\"},{\"content\":\"Successfully downloaded 2574 bytes to example.txt (Content-Type: text/plain; charset=UTF-8)\",\"tool_call_id\":\"call_3eceec4c44ad4007af956277\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/download_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"download the file from https://example-files.online-convert.com/document/txt/example.txt and save it as example.txt\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_506e856f9cbd4a19a541d177\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/download_tool/example.txt\\\",\\\"url\\\":\\\"https://example-files.online-convert.com/document/txt/example.txt\\\"}\",\"name\":\"download\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants to download a file from a URL and save it as example.txt.\"},{\"content\":\"Successfully downloaded 2574 bytes to example.txt (Content-Type: text/plain; charset=UTF-8)\",\"tool_call_id\":\"call_506e856f9cbd4a19a541d177\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -387,27 +563,27 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Download"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Download"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ed"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ed"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" saved"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" saved"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" as"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" as"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"example"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"example"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".txt"},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} - data: {"id":"chatcmpl-24d0696bd856a585bd0a216d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":16,"completion_tokens":11,"total_tokens":7835}} + data: {"id":"chatcmpl-6c6bd5c27eedacaa3068d302","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":80,"completion_tokens":11,"total_tokens":7835}} data: [DONE] @@ -416,4 +592,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.083404s + duration: 3.460239209s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/fetch_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/fetch_tool.yaml index f21defb6df2826dc9de89a6e82b816ca8fde2518..fa8964c630709bd7cb3c5b4eeeb47f19ad327f78 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/fetch_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/fetch_tool.yaml @@ -25,385 +25,165 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" request"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" They"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" want"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fetch"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Fetch"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" check"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" check"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"John"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"John"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'."},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" must"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Need"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Check"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" more"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" than"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".html"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" presence"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" of"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" John"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" One"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Possible"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Check"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Check"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".html"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".html"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" presence"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" of"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" occurrence"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" John"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" of"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" John"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Check"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Check"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" example"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" example"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".html"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".html"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" occurrence"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" presence"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" of"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" of"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" John"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" John"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Check"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"example"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".html"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" total"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"occ"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"urrence"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"of"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"John"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Doe"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Make"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" sure"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fine"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Check"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" example"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".html"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" occurrence"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" of"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" John"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Doe"},"finish_reason":null}]} - - data: {"id":"chatcmpl-0ea6904f89f7f9b1f8ab8eae","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":3,"completion_tokens":197,"total_tokens":424}} + data: {"id":"chatcmpl-636aff44fb03055c1ea3d457","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":51,"completion_tokens":87,"total_tokens":314}} data: [DONE] @@ -412,7 +192,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 697.596792ms + duration: 801.333959ms - id: 1 request: proto: HTTP/1.1 @@ -420,7 +200,7 @@ interactions: proto_minor: 1 content_length: 33725 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/fetch_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"fetch the content from https://example-files.online-convert.com/website/html/example.html and tell me if it contains the word 'John Doe'\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/fetch_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"fetch the content from https://example-files.online-convert.com/website/html/example.html and tell me if it contains the word 'John Doe'\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -437,55 +217,55 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" fetch"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" fetch"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" URL"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" check"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" check"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" if"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" if"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" contains"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" contains"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"John"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"John"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'."},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'."},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_41551da5f8b34a829028b15a","type":"function","index":0,"function":{"name":"fetch","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_927ffc2399f74b52a10e8e52","type":"function","index":0,"function":{"name":"fetch","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"format\":\"text\",\"url\":\"https://example-files.online-convert.com/website/html/example.html\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"format\":\"text\",\"url\":\"https://example-files.online-convert.com/website/html/example.html\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-534faf46f654c6079a08b938","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":15,"completion_tokens":50,"total_tokens":7809}} + data: {"id":"chatcmpl-9c969d015d1eda469a97bacf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":719,"completion_tokens":50,"total_tokens":7809}} data: [DONE] @@ -494,7 +274,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 11.049881958s + duration: 3.823314875s - id: 2 request: proto: HTTP/1.1 @@ -565,7 +345,7 @@ interactions: - text/html; charset=UTF-8 status: 200 OK code: 200 - duration: 142.3385ms + duration: 405.671708ms - id: 3 request: proto: HTTP/1.1 @@ -573,7 +353,7 @@ interactions: proto_minor: 1 content_length: 34254 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/fetch_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"fetch the content from https://example-files.online-convert.com/website/html/example.html and tell me if it contains the word 'John Doe'\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_41551da5f8b34a829028b15a\",\"function\":{\"arguments\":\"{\\\"format\\\":\\\"text\\\",\\\"url\\\":\\\"https://example-files.online-convert.com/website/html/example.html\\\"}\",\"name\":\"fetch\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to fetch content from a URL and check if it contains the word 'John Doe'.\"},{\"content\":\"HTML test file Purpose: Provide example of this file type Document file type: HTML Version: 1.0 Remark:\",\"tool_call_id\":\"call_41551da5f8b34a829028b15a\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/fetch_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"fetch the content from https://example-files.online-convert.com/website/html/example.html and tell me if it contains the word 'John Doe'\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_927ffc2399f74b52a10e8e52\",\"function\":{\"arguments\":\"{\\\"format\\\":\\\"text\\\",\\\"url\\\":\\\"https://example-files.online-convert.com/website/html/example.html\\\"}\",\"name\":\"fetch\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to fetch content from a URL and check if it contains the word 'John Doe'.\"},{\"content\":\"HTML test file Purpose: Provide example of this file type Document file type: HTML Version: 1.0 Remark:\",\"tool_call_id\":\"call_927ffc2399f74b52a10e8e52\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -590,37 +370,37 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"No"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"No"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" fetched"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" fetched"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" does"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" does"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" not"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" not"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" contain"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" contain"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" word"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" word"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"John"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"John"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Doe"},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Doe"},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"'."},"finish_reason":null}]} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"'."},"finish_reason":null}]} - data: {"id":"chatcmpl-f7b2bf895d7376bcf8d5c93f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":8,"completion_tokens":16,"total_tokens":7832}} + data: {"id":"chatcmpl-61f8fe8e1aafe4a79b9447ad","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":72,"completion_tokens":16,"total_tokens":7832}} data: [DONE] @@ -629,4 +409,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 3.944484083s + duration: 3.012608958s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/glob_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/glob_tool.yaml index bb41dabf90f2842f59f7d42867559d2406dae69f..e15683c141aa675877ab1a31eab880a782c805db 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/glob_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/glob_tool.yaml @@ -25,195 +25,199 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Keep"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" find"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Possibly"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" code"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" snippet"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Using"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" locate"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Keep"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Use"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Find"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" list"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" folder"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" craft"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Using"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" locate"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Use"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" list"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Using"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" folder"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" locate"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" inc"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" spaces"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Use"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" list"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Using"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" locate"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" folder"},"finish_reason":null}]} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-216e9f5505919a1c0a6ed41f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":3,"completion_tokens":102,"total_tokens":313}} + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b43fe00f1a52b3fecc7c0543","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":35,"completion_tokens":104,"total_tokens":315}} data: [DONE] @@ -222,7 +226,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 373.542625ms + duration: 716.833959ms - id: 1 request: proto: HTTP/1.1 @@ -230,7 +234,7 @@ interactions: proto_minor: 1 content_length: 33643 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files in the current directory\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files in the current directory\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -247,13 +251,13 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-1b8831d95a34f7b678e905fe","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-1cc33483f7d1f9042df5c7b4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-1b8831d95a34f7b678e905fe","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_d65b8d7f10554c6a9588cd2b","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-1cc33483f7d1f9042df5c7b4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_756ea36486934478a158ca94","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-1b8831d95a34f7b678e905fe","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\",\"pattern\":\"*.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-1cc33483f7d1f9042df5c7b4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\",\"pattern\":\"*.go\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-1b8831d95a34f7b678e905fe","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":62,"completion_tokens":33,"total_tokens":7775}} + data: {"id":"chatcmpl-1cc33483f7d1f9042df5c7b4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":702,"completion_tokens":33,"total_tokens":7775}} data: [DONE] @@ -262,7 +266,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 3.967455291s + duration: 3.384375792s - id: 2 request: proto: HTTP/1.1 @@ -270,7 +274,7 @@ interactions: proto_minor: 1 content_length: 33994 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files in the current directory\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_d65b8d7f10554c6a9588cd2b\",\"function\":{\"arguments\":\"{\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\\",\\\"pattern\\\":\\\"*.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool/main.go\",\"tool_call_id\":\"call_d65b8d7f10554c6a9588cd2b\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files in the current directory\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_756ea36486934478a158ca94\",\"function\":{\"arguments\":\"{\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool\\\",\\\"pattern\\\":\\\"*.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/glob_tool/main.go\",\"tool_call_id\":\"call_756ea36486934478a158ca94\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -287,33 +291,33 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Found"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Found"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `."},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-05b7adc9c33d36b17d9b59cf","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":53,"completion_tokens":14,"total_tokens":7811}} + data: {"id":"chatcmpl-4ffc2a9c94e72823ed7950f6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":117,"completion_tokens":14,"total_tokens":7811}} data: [DONE] @@ -322,4 +326,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 4.257971583s + duration: 3.439591583s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/grep_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/grep_tool.yaml index a76b617609d502cd873d70768018c12a20f1a751..35ab329eace90629e1ee2ac8189a9e907f2ed2a5 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/grep_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/grep_tool.yaml @@ -2,6 +2,86 @@ version: 2 interactions: - id: 0 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 33641 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use grep to search for the word 'package' in go files\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" using"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" grep"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" tool"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_b1755b0046e34decb4062a14","type":"function","index":0,"function":{"name":"grep","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"include\":\"*.go\",\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\",\"pattern\":\"package\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-4970a03e5b3d574139d3ae30","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":703,"completion_tokens":59,"total_tokens":7802}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.903471959s +- id: 1 request: proto: HTTP/1.1 proto_major: 1 @@ -25,213 +105,509 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" generate"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" grep"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" grep"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" maybe"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"rep"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Possibly"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fine"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"rep"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"rep"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" “"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"package"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"”"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" source"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-cfdb60d941abe62528d7086f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":4,"completion_tokens":70,"total_tokens":282}} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 672.711583ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33641 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use grep to search for the word 'package' in go files\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" G"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"r"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"..."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"G"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"rep"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" source"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Gre"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"p"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" use"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"including"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" grep"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" tool"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Actually"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" single"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" word"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" as"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_7a072ff86f5b402f891b4816","type":"function","index":0,"function":{"name":"grep","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'\""},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"include\":\"*.go\",\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\",\"pattern\":\"package\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} - data: {"id":"chatcmpl-8bcd6dd65920726f09a279d9","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":63,"completion_tokens":60,"total_tokens":7803}} + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"including"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"+"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" total"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" so"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" far"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" properly"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"G"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"rep"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \"'"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"package"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"source"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Return"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"rep"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" '"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"package"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"'"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" source"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-32354a7c3f233721ecf44ce4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":36,"completion_tokens":259,"total_tokens":471}} data: [DONE] @@ -240,15 +616,15 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 10.304762834s + duration: 598.054125ms - id: 2 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34175 + content_length: 34174 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use grep to search for the word 'package' in go files\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_7a072ff86f5b402f891b4816\",\"function\":{\"arguments\":\"{\\\"include\\\":\\\"*.go\\\",\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\\",\\\"pattern\\\":\\\"package\\\"}\",\"name\":\"grep\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to use the grep tool to search for the word 'package' in Go files.\"},{\"content\":\"Found 1 matches\\n/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool/main.go:\\n Line 1, Char 1: package main\\n\",\"tool_call_id\":\"call_7a072ff86f5b402f891b4816\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use grep to search for the word 'package' in go files\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_b1755b0046e34decb4062a14\",\"function\":{\"arguments\":\"{\\\"include\\\":\\\"*.go\\\",\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool\\\",\\\"pattern\\\":\\\"package\\\"}\",\"name\":\"grep\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to search for the word 'package' in Go files using the grep tool.\"},{\"content\":\"Found 1 matches\\n/tmp/crush-test/TestCoderAgent/glm-5.1/grep_tool/main.go:\\n Line 1, Char 1: package main\\n\",\"tool_call_id\":\"call_b1755b0046e34decb4062a14\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -265,43 +641,39 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Found"},"finish_reason":null}]} - - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" match"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" match"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" found"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" —"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" contains"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"package"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"package"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} - data: {"id":"chatcmpl-d48cfea83ba8972d9e323702","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":14,"completion_tokens":19,"total_tokens":7841}} + data: {"id":"chatcmpl-86348a2dad25c70a3b77eb78","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":142,"completion_tokens":17,"total_tokens":7839}} data: [DONE] @@ -310,4 +682,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 4.618855916s + duration: 3.811594958s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/ls_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/ls_tool.yaml index fb98b762d0e97bb812dda95ddadf39632a09ea45..5eea472a524216db05ef3d71084c10d1dcf6599b 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/ls_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/ls_tool.yaml @@ -2,6 +2,46 @@ version: 2 interactions: - id: 0 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 33635 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use ls to list the files in the current directory\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-466a263c29c6321b56fb438a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-466a263c29c6321b56fb438a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_2586fc4bcfd24239bcef7ef0","type":"function","index":0,"function":{"name":"ls","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-466a263c29c6321b56fb438a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-466a263c29c6321b56fb438a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":700,"completion_tokens":26,"total_tokens":7766}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.4175995s +- id: 1 request: proto: HTTP/1.1 proto_major: 1 @@ -25,187 +65,399 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" asks"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" given"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" use"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" list"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" list"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Need"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Need"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" output"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"),"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"col"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Files"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Listing"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" good"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" single"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" maybe"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" using"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ls"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-c618e553da7ec0cc185486f0","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":1,"completion_tokens":76,"total_tokens":285}} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Listing"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 658.755125ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33635 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use ls to list the files in the current directory\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Counting"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" letters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" including"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" spaces"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Listing"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Simple"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" request"},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_9c1f574960b54f1b91601bd3","type":"function","index":0,"function":{"name":"ls","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-41d32426ef4f53f94016f3d4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":60,"completion_tokens":28,"total_tokens":7768}} + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"46"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"46"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" char"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Looks"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" One"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Listing"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" current"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" directory"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ls"},"finish_reason":null}]} + + data: {"id":"chatcmpl-2cf9b53fe2aad53110a68ce2","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":33,"completion_tokens":204,"total_tokens":413}} data: [DONE] @@ -214,15 +466,15 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 5.249411208s + duration: 756.2675ms - id: 2 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34020 + content_length: 33983 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use ls to list the files in the current directory\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_9c1f574960b54f1b91601bd3\",\"function\":{\"arguments\":\"{\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\\"}\",\"name\":\"ls\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Simple request\"},{\"content\":\"\\n- /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool/\\n - go.mod\\n - main.go\\n\",\"tool_call_id\":\"call_9c1f574960b54f1b91601bd3\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use ls to list the files in the current directory\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_2586fc4bcfd24239bcef7ef0\",\"function\":{\"arguments\":\"{\\\"path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool\\\"}\",\"name\":\"ls\"},\"type\":\"function\"}],\"role\":\"assistant\"},{\"content\":\"\\n- /tmp/crush-test/TestCoderAgent/glm-5.1/ls_tool/\\n - go.mod\\n - main.go\\n\",\"tool_call_id\":\"call_2586fc4bcfd24239bcef7ef0\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -239,29 +491,33 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Two"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} - data: {"id":"chatcmpl-8b4fac5eeddbcf7f3237757a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":56,"completion_tokens":12,"total_tokens":7812}} + data: {"id":"chatcmpl-b8514ff71d680adaf2f0e430","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":120,"completion_tokens":14,"total_tokens":7814}} data: [DONE] @@ -270,4 +526,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.308737667s + duration: 3.19619575s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/multiedit_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/multiedit_tool.yaml index a38dc1d5f536c9f7ed7ad45738606e23ad8df024..b0c61ea39dbcef680eb1ad42ac23431c4b7058b5 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/multiedit_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/multiedit_tool.yaml @@ -2,6 +2,114 @@ version: 2 interactions: - id: 0 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 33721 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Let"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" first"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" see"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" its"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" contents"},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_b2789b48d2ec41da9f61a574","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-54dc2649ab88ccb894885824","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":725,"completion_tokens":45,"total_tokens":7810}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.823122208s +- id: 1 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 34251 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_b2789b48d2ec41da9f61a574\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me first read the main.go file to see its current contents.\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_b2789b48d2ec41da9f61a574\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-517acaae8a6b55fa3c6a5333","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-517acaae8a6b55fa3c6a5333","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_b3accc0a7eaa4756b961e075","type":"function","index":0,"function":{"name":"multiedit","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-517acaae8a6b55fa3c6a5333","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"edits\":[{\"new_string\": \"\\t// Greeting\\n\\tfmt.Println(\\\"Hello, Crush!\\\")\", \"old_string\": \"\\tfmt.Println(\\\"Hello, World!\\\")\"}],\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-517acaae8a6b55fa3c6a5333","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":108,"completion_tokens":76,"total_tokens":7928}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 4.295830959s +- id: 2 request: proto: HTTP/1.1 proto_major: 1 @@ -25,751 +133,779 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" request"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Language"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" English"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Using"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Change"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"!"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"!"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" within"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Change"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Could"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Change"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" inc"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?)"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Actually"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"M"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"...)."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Then"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" total"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"!("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"g"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"reeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" exactly"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Actually"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" precisely"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" including"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" spaces"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" string"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Change"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Write"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" out"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" C"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"h"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" within"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" limit"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"n"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"g"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Rule"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"e"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" says"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" letters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" It"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" says"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" H"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Do"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"e"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" not"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" use"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"l"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"l"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"o"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" cannot"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" use"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" W"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Remove"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"o"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"r"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"l"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"d"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"o"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" H"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"e"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"l"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"l"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"o"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" C"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"r"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"u"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"s"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"h"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"g"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"n"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"reeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"d"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"d"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" =\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"d"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":";"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" c"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"o"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"m"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"45"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"m"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"46"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"e"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"47"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"n"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"48"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"t"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" But"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" maybe"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" better"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Replace"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"=\u003e"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Mult"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"ied"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Replace"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"it"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" change"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" add"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5a839663827ce384987ebda4","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":7,"completion_tokens":328,"total_tokens":559}} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 395.249792ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33721 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Let"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" first"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" World"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" see"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" its"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" contents"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_3fc8df8501a54bdb8a533d23","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-d26fcae8158f8db937b68814","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":21,"completion_tokens":43,"total_tokens":7808}} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 10.044032875s -- id: 2 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 34243 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_3fc8df8501a54bdb8a533d23\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me first read the file to see its current contents.\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_3fc8df8501a54bdb8a533d23\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-fe2d485c9056c0fc0c37fa57","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-fe2d485c9056c0fc0c37fa57","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_162949daf153494e86785c95","type":"function","index":0,"function":{"name":"multiedit","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-fe2d485c9056c0fc0c37fa57","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"edits\":[{\"new_string\": \"Hello, Crush!\", \"old_string\": \"Hello, World!\"}, {\"new_string\": \"\\t// Greeting\\n\\tfmt.Println(\\\"Hello, Crush!\\\")\", \"old_string\": \"\\tfmt.Println(\\\"Hello, Crush!\\\")\"}],\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} - data: {"id":"chatcmpl-fe2d485c9056c0fc0c37fa57","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":44,"completion_tokens":95,"total_tokens":7947}} + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" comment"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"48"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Under"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" output"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Replace"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" World"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" comment"},"finish_reason":null}]} + + data: {"id":"chatcmpl-18ba696a53142f8fceb16d9d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":55,"completion_tokens":394,"total_tokens":625}} data: [DONE] @@ -778,15 +914,15 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 5.825693125s + duration: 706.161541ms - id: 3 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34892 + content_length: 34828 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_3fc8df8501a54bdb8a533d23\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me first read the file to see its current contents.\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_3fc8df8501a54bdb8a533d23\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_162949daf153494e86785c95\",\"function\":{\"arguments\":\"{\\\"edits\\\":[{\\\"new_string\\\": \\\"Hello, Crush!\\\", \\\"old_string\\\": \\\"Hello, World!\\\"}, {\\\"new_string\\\": \\\"\\\\t// Greeting\\\\n\\\\tfmt.Println(\\\\\\\"Hello, Crush!\\\\\\\")\\\", \\\"old_string\\\": \\\"\\\\tfmt.Println(\\\\\\\"Hello, Crush!\\\\\\\")\\\"}],\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"multiedit\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\nApplied 2 edits to file: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\n\\n\",\"tool_call_id\":\"call_162949daf153494e86785c95\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use multiedit to change 'Hello, World!' to 'Hello, Crush!' and add a comment '// Greeting' above the fmt.Println line in main.go\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_b2789b48d2ec41da9f61a574\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me first read the main.go file to see its current contents.\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_b2789b48d2ec41da9f61a574\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_b3accc0a7eaa4756b961e075\",\"function\":{\"arguments\":\"{\\\"edits\\\":[{\\\"new_string\\\": \\\"\\\\t// Greeting\\\\n\\\\tfmt.Println(\\\\\\\"Hello, Crush!\\\\\\\")\\\", \\\"old_string\\\": \\\"\\\\tfmt.Println(\\\\\\\"Hello, World!\\\\\\\")\\\"}],\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\\"}\",\"name\":\"multiedit\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\nApplied 1 edits to file: /tmp/crush-test/TestCoderAgent/glm-5.1/multiedit_tool/main.go\\n\\n\",\"tool_call_id\":\"call_b3accc0a7eaa4756b961e075\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -803,39 +939,49 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Done"},"finish_reason":null}]} + + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Changed"},"finish_reason":null}]} + + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Done"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" string"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Updated"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" added"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" string"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"//"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" G"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" added"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"reeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" comment"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" comment"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" above"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" above"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"fmt"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"fmt"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".Println"},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".Println"},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} - data: {"id":"chatcmpl-3326c09673a979871ae3aa86","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":114,"completion_tokens":17,"total_tokens":8003}} + data: {"id":"chatcmpl-291b41722ec3e5d4a675d235","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":159,"completion_tokens":22,"total_tokens":7989}} data: [DONE] @@ -844,4 +990,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.399328375s + duration: 3.39534875s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/parallel_tool_calls.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/parallel_tool_calls.yaml index ff526a9794312f2f877966ce8e271496650f330e..05c34f04206cf012cd5a846020476f46d78d1906 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/parallel_tool_calls.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/parallel_tool_calls.yaml @@ -25,357 +25,479 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" generate"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" request"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Parallel"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ≤"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Keep"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Parallel"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Content"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" find"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" all"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" use"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Parallel"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" list"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" very"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" important"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" you"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" run"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" both"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" tool"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" calls"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Parallel"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" parallel"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" could"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Parallel"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Parallel"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"for"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" files"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" P"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"files"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"8"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" g"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"and"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" b"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" f"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ."},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" g"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"("},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"21"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" f"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" okay"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"25"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Parallel"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" glob"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ."},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" n"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" d"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-b682e93c4db04e3e10b56efc","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":4,"completion_tokens":183,"total_tokens":411}} + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"31"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" l"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" d"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" i"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" e"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" c"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"39"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" t"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" o"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" r"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"42"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" y"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"43"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" One"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"colon"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fits"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Return"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Parallel"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" glob"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ."},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" files"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" ls"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" directory"},"finish_reason":null}]} + + data: {"id":"chatcmpl-fd45114ba4070538ac1dc9db","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":36,"completion_tokens":244,"total_tokens":472}} data: [DONE] @@ -384,7 +506,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 788.312459ms + duration: 534.428042ms - id: 1 request: proto: HTTP/1.1 @@ -392,7 +514,7 @@ interactions: proto_minor: 1 content_length: 33732 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files and use ls to list the current directory, it is very important that you run both tool calls in parallel\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files and use ls to list the current directory, it is very important that you run both tool calls in parallel\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -409,17 +531,43 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" run"},"finish_reason":null}]} - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_1d40512641f0427abfe1c16a","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" both"},"finish_reason":null}]} - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"pattern\":\"**/*.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" glob"},"finish_reason":null}]} - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_2283f0d4c8484e8fae126ee9","type":"function","index":1,"function":{"name":"ls","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":1,"function":{"arguments":"{}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" ls"},"finish_reason":null}]} - data: {"id":"chatcmpl-1ad3bedf586346b742acf111","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":16,"completion_tokens":16,"total_tokens":7776}} + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" parallel"},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_5f2279f0e7244e2d9fb544b2","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"pattern\":\"**/*.go\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_c586cf0dd7ad4bf0a81467e1","type":"function","index":1,"function":{"name":"ls","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":1,"function":{"arguments":"{}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-76e19368c1ae2b71a859760b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":720,"completion_tokens":29,"total_tokens":7789}} data: [DONE] @@ -428,15 +576,15 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 6.799006417s + duration: 3.448828125s - id: 2 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34299 + content_length: 34376 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files and use ls to list the current directory, it is very important that you run both tool calls in parallel\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_1d40512641f0427abfe1c16a\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/*.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"},{\"id\":\"call_2283f0d4c8484e8fae126ee9\",\"function\":{\"arguments\":\"{}\",\"name\":\"ls\"},\"type\":\"function\"}],\"role\":\"assistant\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls/main.go\",\"tool_call_id\":\"call_1d40512641f0427abfe1c16a\",\"role\":\"tool\"},{\"content\":\"\\n- /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls/\\n - go.mod\\n - main.go\\n\",\"tool_call_id\":\"call_2283f0d4c8484e8fae126ee9\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use glob to find all .go files and use ls to list the current directory, it is very important that you run both tool calls in parallel\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_5f2279f0e7244e2d9fb544b2\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/*.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"},{\"id\":\"call_c586cf0dd7ad4bf0a81467e1\",\"function\":{\"arguments\":\"{}\",\"name\":\"ls\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to run both glob and ls in parallel.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls/main.go\",\"tool_call_id\":\"call_5f2279f0e7244e2d9fb544b2\",\"role\":\"tool\"},{\"content\":\"\\n- /tmp/crush-test/TestCoderAgent/glm-5.1/parallel_tool_calls/\\n - go.mod\\n - main.go\\n\",\"tool_call_id\":\"call_c586cf0dd7ad4bf0a81467e1\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -453,67 +601,71 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**"},"finish_reason":null}]} + + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"lob"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"lob"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Found"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `."},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `."},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" found"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" —"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`\n\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Directory"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" listing"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" listing"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Contains"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`,"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" and"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-8e05ed9bad65b38f2bbb5ff6","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":24,"completion_tokens":31,"total_tokens":7863}} + data: {"id":"chatcmpl-b5c07e7fd88fa7ed34fd187f","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":88,"completion_tokens":33,"total_tokens":7865}} data: [DONE] @@ -522,4 +674,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.634782666s + duration: 3.486510084s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/read_a_file.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/read_a_file.yaml index 34b2863d7230dff167aaefb1cebba5c131fea3e5..2ae9e2d285ef53cdcf4b51916b387d10d17bb7d3 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/read_a_file.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/read_a_file.yaml @@ -25,95 +25,101 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Read"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" mod"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Read"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Probably"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Reading"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" English"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" File"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Keep"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" under"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Viewing"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Should"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Module"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" File"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Read"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" fine"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" go"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Viewing"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" File"},"finish_reason":null}]} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-98d249b7bf501e68c95ad23d","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":11,"completion_tokens":52,"total_tokens":255}} + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" module"},"finish_reason":null}]} + + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" file"},"finish_reason":null}]} + + data: {"id":"chatcmpl-30ece0378efe0089c19682b8","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":27,"completion_tokens":55,"total_tokens":258}} data: [DONE] @@ -122,7 +128,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 424.473125ms + duration: 626.797666ms - id: 1 request: proto: HTTP/1.1 @@ -130,7 +136,7 @@ interactions: proto_minor: 1 content_length: 33605 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Read the go mod\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Read the go mod\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -147,43 +153,47 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" go"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".mod"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".mod"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Let"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" current"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" directory"},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" find"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" it"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_3e347af74bdd4b0dae2e851f","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" first"},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-e90396f8ffbdb35296103d22","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":54,"completion_tokens":44,"total_tokens":7778}} + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_369b0f9d79344c8f81cd6e25","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"pattern\":\"**/go.mod\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-9ebbe210b1528cc5f0518646","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":694,"completion_tokens":31,"total_tokens":7765}} data: [DONE] @@ -192,15 +202,55 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 5.848403958s + duration: 3.653444417s - id: 2 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34064 + content_length: 33985 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Read the go mod\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_369b0f9d79344c8f81cd6e25\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/go.mod\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to read a go.mod file. Let me find it first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\",\"tool_call_id\":\"call_369b0f9d79344c8f81cd6e25\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-cdc4012dff77ab9c4d4cacc4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-cdc4012dff77ab9c4d4cacc4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_998b26c587834e75acb5f123","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-cdc4012dff77ab9c4d4cacc4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-cdc4012dff77ab9c4d4cacc4","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":90,"completion_tokens":29,"total_tokens":7799}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.295979958s +- id: 3 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 34377 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Read the go mod\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_3e347af74bdd4b0dae2e851f\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to read the go.mod file in the current directory.\"},{\"content\":\"\\n 1|module example.com/testproject\\n 2|\\n 3|go 1.23\\n\\n\",\"tool_call_id\":\"call_3e347af74bdd4b0dae2e851f\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Read the go mod\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_369b0f9d79344c8f81cd6e25\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/go.mod\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants me to read a go.mod file. Let me find it first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\",\"tool_call_id\":\"call_369b0f9d79344c8f81cd6e25\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_998b26c587834e75acb5f123\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/read_a_file/go.mod\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\n 1|module example.com/testproject\\n 2|\\n 3|go 1.23\\n\\n\",\"tool_call_id\":\"call_998b26c587834e75acb5f123\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -217,39 +267,61 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"The"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".mod"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" defines"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":\n\n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" **"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Module"},"finish_reason":null}]} + + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Module"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"example"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".com"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"example"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/test"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".com"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"project"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/test"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"project"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`,"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" **"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" version"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" version"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"**:"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"23"},"finish_reason":null}]} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-7507ade6feb466a87a289dfb","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":117,"completion_tokens":17,"total_tokens":7814}} + data: {"id":"chatcmpl-acf305c37f7aff09217e85da","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":89,"completion_tokens":28,"total_tokens":7861}} data: [DONE] @@ -258,4 +330,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.068550917s + duration: 3.603572084s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/simple_test.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/simple_test.yaml index bf1a3a0eb5918fa0f644cbd7c59c71a527daafbb..43c2ef70079818a45976f636c3fd04f4d7036be0 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/simple_test.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/simple_test.yaml @@ -25,151 +25,143 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" says"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" asks"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" They"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" want"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" The"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" just"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" We"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" must"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" output"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" more"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" than"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" could"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summary"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" of"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user's"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Or"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" message"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Simple"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" something"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Use"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" appropriate"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Provide"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Return"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" just"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" I'll"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" output"},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Greeting"},"finish_reason":null}]} - - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Greeting"},"finish_reason":null}]} - - data: {"id":"chatcmpl-9b6c8c4fb96fca89f847f69c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":8,"completion_tokens":80,"total_tokens":280}} + data: {"id":"chatcmpl-a2f612ae965c35df07c936e7","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":152,"completion_tokens":76,"total_tokens":276}} data: [DONE] @@ -178,7 +170,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 626.287542ms + duration: 1.432785917s - id: 1 request: proto: HTTP/1.1 @@ -186,7 +178,7 @@ interactions: proto_minor: 1 content_length: 33595 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/simple_test\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Hello\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/simple_test\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"Hello\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -203,61 +195,57 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" just"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" said"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" is"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" simply"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" I"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" respond"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" I"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" briefly"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" respond"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" conc"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" conc"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"is"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"is"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"ely"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"ely"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Hi"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" How"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" How"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" can"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" can"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" I"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" I"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" help"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" help"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" you"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" you"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" today"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" today"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"?"},"finish_reason":null}]} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-8f82f3a7d2915d3316317d11","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":50,"completion_tokens":27,"total_tokens":7757}} + data: {"id":"chatcmpl-2d444759c1a25aa3a313fbd2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":7730,"completion_tokens":25,"total_tokens":7755}} data: [DONE] @@ -266,4 +254,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 6.1273815s + duration: 4.787780708s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/sourcegraph_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/sourcegraph_tool.yaml index 880ba37f29f217c8103a2ed812cd3286ec48edc3..70cf364161079c467f6e7221581cf06f847ef86f 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/sourcegraph_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/sourcegraph_tool.yaml @@ -6,9 +6,9 @@ interactions: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 819 + content_length: 33655 host: "" - body: '{"messages":[{"content":"You will generate a short title based on the first message a user begins a conversation with.\n\n\n- Keep the title in the same language that the user wrote their message in.\n- Ensure it is not more than 50 characters long.\n- The title should be a summary of the user''s message.\n- It should be one line long.\n- Do not use quotes or colons.\n- The entire text you return will be used as the title.\n- Never return anything that is more than one sentence (one line) long.\n\n\n /no_think","role":"system"},{"content":"Generate a concise title for the following content:\n\nuse sourcegraph to search for ''func main'' in Go repositories\n \n\n","role":"user"}],"model":"gpt-oss-120b","max_completion_tokens":40,"stream_options":{"include_usage":true},"stream":true}' + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/sourcegraph_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use sourcegraph to search for 'func main' in Go repositories\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -25,633 +25,405 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" asks"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":\n\n"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" source"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"func"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repositories"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"\n\n"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" produce"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"max"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summar"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"izing"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Something"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" like"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Searching"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repos"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" within"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Searching"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repos"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" including"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" spaces"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" S"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"earch"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ing"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" repositories"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" using"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" tool"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_55e4aed5dc874cca8e0a17fb","type":"function","index":0,"function":{"name":"sourcegraph","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"query\":\"func main lang:go\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-1a03b5482191d75bd53c970b","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":703,"completion_tokens":35,"total_tokens":7778}} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"26"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: [DONE] - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.972791416s +- id: 1 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 819 + host: "" + body: '{"messages":[{"content":"You will generate a short title based on the first message a user begins a conversation with.\n\n\n- Keep the title in the same language that the user wrote their message in.\n- Ensure it is not more than 50 characters long.\n- The title should be a summary of the user''s message.\n- It should be one line long.\n- Do not use quotes or colons.\n- The entire text you return will be used as the title.\n- Never return anything that is more than one sentence (one line) long.\n\n\n /no_think","role":"system"},{"content":"Generate a concise title for the following content:\n\nuse sourcegraph to search for ''func main'' in Go repositories\n \n\n","role":"user"}],"model":"gpt-oss-120b","max_completion_tokens":40,"stream_options":{"include_usage":true},"stream":true}' + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repos"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repositories"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" must"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ≤"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"41"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"52"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Too"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" long"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Possible"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"52"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Need"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Searching"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Re"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"pos"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Maybe"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Search"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repos"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" via"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Searching"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Re"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Search"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"pos"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Searching"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"9"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"16"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"12"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"13"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Re"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"pos"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"20"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"19"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"22"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" repos"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"27"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"28"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" via"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"32"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"33"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"37"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"38"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" total"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"44"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"11"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ok"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Total"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" single"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"49"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" okay"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Return"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Provide"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Searching"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Search"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" func"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Re"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"pos"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" func"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" repos"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" via"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Source"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" Source"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"graph"},"finish_reason":null}]} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"graph"},"finish_reason":null}]} - data: {"id":"chatcmpl-a79ccf0f2569b098b9e53a5f","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":4,"completion_tokens":321,"total_tokens":533}} + data: {"id":"chatcmpl-56ce039883eb4a6456955034","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":36,"completion_tokens":168,"total_tokens":380}} data: [DONE] @@ -660,85 +432,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 472.076292ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33655 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/sourcegraph_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use sourcegraph to search for 'func main' in Go repositories\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" search"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" '"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"func"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Go"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" repositories"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" using"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" Source"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"graph"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" tool"},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_cab40f2307014839bb51be75","type":"function","index":0,"function":{"name":"sourcegraph","arguments":""}}]},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"count\":10,\"query\":\"func main lang:go\"}"}}]},"finish_reason":null}]} - - data: {"id":"chatcmpl-19ed4928016cf20e8b8f9035","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":63,"completion_tokens":41,"total_tokens":7784}} - - data: [DONE] - - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 9.831210208s + duration: 600.888958ms - id: 2 request: proto: HTTP/1.1 @@ -760,21 +454,21 @@ interactions: proto_minor: 0 content_length: -1 uncompressed: true - body: '{"data":{"search":{"results":{"matchCount":30,"limitHit":true,"resultCount":30,"approximateResultCount":"30+","missing":[],"timedout":[],"indexUnavailable":false,"results":[{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/01_before-func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/01_before-func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc greeting() {\n\tfmt.Println(\"Hello world!\")\n}\n\nfunc main() {\n\tgreeting()\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":8,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func greeting() {","lineNumber":4,"offsetAndLengths":[[0,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/02_func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/02_func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc main() {\n\n\tgreeting := func() {\n\t\tfmt.Println(\"Hello world!\")\n\t}\n\n\tgreeting()\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":4,"offsetAndLengths":[[0,9]]},{"preview":"\tgreeting := func() {","lineNumber":6,"offsetAndLengths":[[13,4]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/03_func-expression_shows-type/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/03_func-expression_shows-type/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc main() {\n\n\tgreeting := func() {\n\t\tfmt.Println(\"Hello world!\")\n\t}\n\n\tgreeting()\n\tfmt.Printf(\"%T\\n\", greeting)\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":4,"offsetAndLengths":[[0,9]]},{"preview":"\tgreeting := func() {","lineNumber":6,"offsetAndLengths":[[13,4]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/04_another-way_func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/04_another-way_func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc makeGreeter() func() string {\n\treturn func() string {\n\t\treturn \"Hello world!\"\n\t}\n}\n\nfunc main() {\n\tgreet := makeGreeter()\n\tfmt.Println(greet())\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":10,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func makeGreeter() func() string {","lineNumber":4,"offsetAndLengths":[[0,4],[19,4]]},{"preview":"\treturn func() string {","lineNumber":5,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc makeGreeter() func() string {\n\treturn func() string {\n\t\treturn \"Hello world!\"\n\t}\n}\n\nfunc main() {\n\tgreet := makeGreeter()\n\tfmt.Println(greet())\n\tfmt.Printf(\"%T\\n\", greet)\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":10,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func makeGreeter() func() string {","lineNumber":4,"offsetAndLengths":[[0,4],[19,4]]},{"preview":"\treturn func() string {","lineNumber":5,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/webtao520/golang_project"},"file":{"path":"golang项目实战架构/day03/panic_recover/main01.go","url":"/r/github.com/webtao520/golang_project/-/blob/golang%E9%A1%B9%E7%9B%AE%E5%AE%9E%E6%88%98%E6%9E%B6%E6%9E%84/day03/panic_recover/main01.go","content":"package main \n\n\nimport (\n\t\"fmt\"\n)\n\nfunc funcA() {\n\tfmt.Println(\"func A\")\n}\n\nfunc funcB() {\n\tpanic(\"panic in B\")\n}\n\nfunc funcC() {\n\tfmt.Println(\"func C\")\n}\n\n// 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\nfunc main() {\n\tfuncA()\n\tfuncB()\n\tfuncC()\n}\n\n/**\nPS D:\\goLang\\github\\golang_project\\golang项目实战架构\\day03\\panic_recover\u003e go run .\\main01.go\nfunc A\npanic: panic in B\n\ngoroutine 1 [running]:\nmain.funcB(...)\n D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13\nmain.main()\n D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\nexit status 2\n*/"},"lineMatches":[{"preview":"// 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了","lineNumber":19,"offsetAndLengths":[[9,4]]},{"preview":"func main() {","lineNumber":20,"offsetAndLengths":[[0,9]]},{"preview":"\tfuncA()","lineNumber":21,"offsetAndLengths":[[1,4]]},{"preview":"\tfuncB()","lineNumber":22,"offsetAndLengths":[[1,4]]},{"preview":"\tfuncC()","lineNumber":23,"offsetAndLengths":[[1,4]]},{"preview":"main.funcB(...)","lineNumber":32,"offsetAndLengths":[[0,4],[5,4]]},{"preview":" D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13","lineNumber":33,"offsetAndLengths":[[73,4]]},{"preview":"main.main()","lineNumber":34,"offsetAndLengths":[[0,4],[5,4]]},{"preview":" D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d","lineNumber":35,"offsetAndLengths":[[73,4]]}]}]}}}}' + body: '{"data":{"search":{"results":{"matchCount":30,"limitHit":true,"resultCount":30,"approximateResultCount":"30+","missing":[],"timedout":[],"indexUnavailable":false,"results":[{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/01_before-func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/01_before-func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc greeting() {\n\tfmt.Println(\"Hello world!\")\n}\n\nfunc main() {\n\tgreeting()\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":8,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func greeting() {","lineNumber":4,"offsetAndLengths":[[0,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/02_func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/02_func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc main() {\n\n\tgreeting := func() {\n\t\tfmt.Println(\"Hello world!\")\n\t}\n\n\tgreeting()\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":4,"offsetAndLengths":[[0,9]]},{"preview":"\tgreeting := func() {","lineNumber":6,"offsetAndLengths":[[13,4]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/03_func-expression_shows-type/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/03_func-expression_shows-type/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc main() {\n\n\tgreeting := func() {\n\t\tfmt.Println(\"Hello world!\")\n\t}\n\n\tgreeting()\n\tfmt.Printf(\"%T\\n\", greeting)\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":4,"offsetAndLengths":[[0,9]]},{"preview":"\tgreeting := func() {","lineNumber":6,"offsetAndLengths":[[13,4]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/04_another-way_func-expression/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/04_another-way_func-expression/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc makeGreeter() func() string {\n\treturn func() string {\n\t\treturn \"Hello world!\"\n\t}\n}\n\nfunc main() {\n\tgreet := makeGreeter()\n\tfmt.Println(greet())\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":10,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func makeGreeter() func() string {","lineNumber":4,"offsetAndLengths":[[0,4],[19,4]]},{"preview":"\treturn func() string {","lineNumber":5,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/GoesToEleven/GolangTraining"},"file":{"path":"14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go","url":"/r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go","content":"package main\n\nimport \"fmt\"\n\nfunc makeGreeter() func() string {\n\treturn func() string {\n\t\treturn \"Hello world!\"\n\t}\n}\n\nfunc main() {\n\tgreet := makeGreeter()\n\tfmt.Println(greet())\n\tfmt.Printf(\"%T\\n\", greet)\n}\n"},"lineMatches":[{"preview":"func main() {","lineNumber":10,"offsetAndLengths":[[0,9]]},{"preview":"package main","lineNumber":0,"offsetAndLengths":[[8,4]]},{"preview":"func makeGreeter() func() string {","lineNumber":4,"offsetAndLengths":[[0,4],[19,4]]},{"preview":"\treturn func() string {","lineNumber":5,"offsetAndLengths":[[8,4]]}]},{"__typename":"FileMatch","repository":{"name":"github.com/vitorbritto/dexter-momentum"},"file":{"path":"go/concepts/10-functions/stack-function/main.go","url":"/r/github.com/vitorbritto/dexter-momentum/-/blob/go/concepts/10-functions/stack-function/main.go","content":"package main\n\nimport \"runtime/debug\"\n\nfunc f3() {\n\tdebug.PrintStack()\n}\n\nfunc f2() {\n\tf3()\n}\n\nfunc f1() {\n\tf2()\n}\n\nfunc main() {\n\tf1()\n}\n\n// Output:\n// goroutine 1 [running]:\n// main.f3()\n// \t/Users/jason/go/src/stack-function/main.go:3 +0x11\n// main.f2()\n// \t/Users/jason/go/src/stack-function/main.go:7 +0x11\n// main.f1()\n// \t/Users/jason/go/src/stack-function/main.go:11 +0x11\n// main.main()\n// \t/Users/jason/go/src/stack-function/main.go:15 +0x11\n"},"lineMatches":[{"preview":"func main() {","lineNumber":16,"offsetAndLengths":[[0,9]]},{"preview":"// main.f3()","lineNumber":22,"offsetAndLengths":[[3,4]]},{"preview":"// \t/Users/jason/go/src/stack-function/main.go:3 +0x11","lineNumber":23,"offsetAndLengths":[[30,4],[39,4]]},{"preview":"// main.f2()","lineNumber":24,"offsetAndLengths":[[3,4]]},{"preview":"// \t/Users/jason/go/src/stack-function/main.go:7 +0x11","lineNumber":25,"offsetAndLengths":[[30,4],[39,4]]},{"preview":"// main.f1()","lineNumber":26,"offsetAndLengths":[[3,4]]},{"preview":"// \t/Users/jason/go/src/stack-function/main.go:11 +0x11","lineNumber":27,"offsetAndLengths":[[30,4],[39,4]]},{"preview":"// main.main()","lineNumber":28,"offsetAndLengths":[[3,4]]}]}]}}}}' headers: Content-Type: - application/json status: 200 OK code: 200 - duration: 1.721604416s + duration: 1.355889541s - id: 3 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 44209 + content_length: 43109 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/sourcegraph_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use sourcegraph to search for 'func main' in Go repositories\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_cab40f2307014839bb51be75\",\"function\":{\"arguments\":\"{\\\"count\\\":10,\\\"query\\\":\\\"func main lang:go\\\"}\",\"name\":\"sourcegraph\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants to search for 'func main' in Go repositories using the Sourcegraph tool.\"},{\"content\":\"# Sourcegraph Search Results\\n\\nFound 30 matches across 30 results\\n(Result limit reached, try a more specific query)\\n\\n## Result 1: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/01_before-func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/01_before-func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| func main() {\\n9| func main() {\\n10| \\tgreeting()\\n11| }\\n12| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| \\n9| func main() {\\n10| \\tgreeting()\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func greeting() {\\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| \\n9| func main() {\\n10| \\tgreeting()\\n11| }\\n12| \\n```\\n\\n## Result 2: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/02_func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/02_func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func main() {\\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| }\\n13| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\tgreeting := func() {\\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| }\\n13| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n```\\n\\n## Result 3: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/03_func-expression_shows-type/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/03_func-expression_shows-type/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func main() {\\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| \\tfmt.Printf(\\\"%T\\\\n\\\", greeting)\\n13| }\\n14| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\tgreeting := func() {\\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| \\tfmt.Printf(\\\"%T\\\\n\\\", greeting)\\n13| }\\n14| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n```\\n\\n## Result 4: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/04_another-way_func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/04_another-way_func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| func main() {\\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n15| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func makeGreeter() func() string {\\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| \\treturn func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n15| \\n```\\n\\n## Result 5: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| func main() {\\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n15| }\\n16| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func makeGreeter() func() string {\\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| \\treturn func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n15| }\\n```\\n\\n## Result 6: github.com/webtao520/golang_project/golang项目实战架构/day03/panic_recover/main01.go\\n\\nURL: /r/github.com/webtao520/golang_project/-/blob/golang%E9%A1%B9%E7%9B%AE%E5%AE%9E%E6%88%98%E6%9E%B6%E6%9E%84/day03/panic_recover/main01.go\\n\\n```\\n9| \\tfmt.Println(\\\"func A\\\")\\n10| }\\n11| \\n12| func funcB() {\\n13| \\tpanic(\\\"panic in B\\\")\\n14| }\\n15| \\n16| func funcC() {\\n17| \\tfmt.Println(\\\"func C\\\")\\n18| }\\n19| // 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\\n20| // 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\\n21| func main() {\\n22| \\tfuncA()\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n```\\n\\n```\\n10| }\\n11| \\n12| func funcB() {\\n13| \\tpanic(\\\"panic in B\\\")\\n14| }\\n15| \\n16| func funcC() {\\n17| \\tfmt.Println(\\\"func C\\\")\\n18| }\\n19| \\n20| func main() {\\n21| func main() {\\n22| \\tfuncA()\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n```\\n\\n```\\n11| \\n12| func funcB() {\\n13| \\tpanic(\\\"panic in B\\\")\\n14| }\\n15| \\n16| func funcC() {\\n17| \\tfmt.Println(\\\"func C\\\")\\n18| }\\n19| \\n20| // 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\\n21| \\tfuncA()\\n22| \\tfuncA()\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n```\\n\\n```\\n12| func funcB() {\\n13| \\tpanic(\\\"panic in B\\\")\\n14| }\\n15| \\n16| func funcC() {\\n17| \\tfmt.Println(\\\"func C\\\")\\n18| }\\n19| \\n20| // 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\\n21| func main() {\\n22| \\tfuncB()\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| goroutine 1 [running]:\\n```\\n\\n```\\n13| \\tpanic(\\\"panic in B\\\")\\n14| }\\n15| \\n16| func funcC() {\\n17| \\tfmt.Println(\\\"func C\\\")\\n18| }\\n19| \\n20| // 程序运行期间funcB中引发了panic导致程序崩溃，异常退出了\\n21| func main() {\\n22| \\tfuncA()\\n23| \\tfuncC()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| goroutine 1 [running]:\\n33| main.funcB(...)\\n```\\n\\n```\\n22| \\tfuncA()\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| main.funcB(...)\\n33| main.funcB(...)\\n34| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13\\n35| main.main()\\n36| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\\n37| exit status 2\\n38| */\\n```\\n\\n```\\n23| \\tfuncB()\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| goroutine 1 [running]:\\n33| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13\\n34| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13\\n35| main.main()\\n36| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\\n37| exit status 2\\n38| */\\n```\\n\\n```\\n24| \\tfuncC()\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| goroutine 1 [running]:\\n33| main.funcB(...)\\n34| main.main()\\n35| main.main()\\n36| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\\n37| exit status 2\\n38| */\\n```\\n\\n```\\n25| }\\n26| \\n27| /**\\n28| PS D:\\\\goLang\\\\github\\\\golang_project\\\\golang项目实战架构\\\\day03\\\\panic_recover> go run .\\\\main01.go\\n29| func A\\n30| panic: panic in B\\n31| \\n32| goroutine 1 [running]:\\n33| main.funcB(...)\\n34| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:13\\n35| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\\n36| D:/goLang/github/golang_project/golang项目实战架构/day03/panic_recover/main01.go:23 +0x9d\\n37| exit status 2\\n38| */\\n```\\n\\n\",\"tool_call_id\":\"call_cab40f2307014839bb51be75\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/sourcegraph_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use sourcegraph to search for 'func main' in Go repositories\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_55e4aed5dc874cca8e0a17fb\",\"function\":{\"arguments\":\"{\\\"query\\\":\\\"func main lang:go\\\"}\",\"name\":\"sourcegraph\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"The user wants to search for 'func main' in Go repositories using the Sourcegraph tool.\"},{\"content\":\"# Sourcegraph Search Results\\n\\nFound 30 matches across 30 results\\n(Result limit reached, try a more specific query)\\n\\n## Result 1: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/01_before-func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/01_before-func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| func main() {\\n9| func main() {\\n10| \\tgreeting()\\n11| }\\n12| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| \\n9| func main() {\\n10| \\tgreeting()\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func greeting() {\\n5| func greeting() {\\n6| \\tfmt.Println(\\\"Hello world!\\\")\\n7| }\\n8| \\n9| func main() {\\n10| \\tgreeting()\\n11| }\\n12| \\n```\\n\\n## Result 2: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/02_func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/02_func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func main() {\\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| }\\n13| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\tgreeting := func() {\\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| }\\n13| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n```\\n\\n## Result 3: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/03_func-expression_shows-type/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/03_func-expression_shows-type/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func main() {\\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| \\tfmt.Printf(\\\"%T\\\\n\\\", greeting)\\n13| }\\n14| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\tgreeting := func() {\\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n11| \\tgreeting()\\n12| \\tfmt.Printf(\\\"%T\\\\n\\\", greeting)\\n13| }\\n14| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func main() {\\n6| \\n7| \\tgreeting := func() {\\n8| \\t\\tfmt.Println(\\\"Hello world!\\\")\\n9| \\t}\\n10| \\n```\\n\\n## Result 4: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/04_another-way_func-expression/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/04_another-way_func-expression/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| func main() {\\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n15| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func makeGreeter() func() string {\\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| \\treturn func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| }\\n15| \\n```\\n\\n## Result 5: github.com/GoesToEleven/GolangTraining/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go\\n\\nURL: /r/github.com/GoesToEleven/GolangTraining/-/blob/14_functions/10_func-expression/05_another-way_func-expression_shows-type/main.go\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| func main() {\\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n15| }\\n16| \\n```\\n\\n```\\n0| package main\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| func makeGreeter() func() string {\\n5| func makeGreeter() func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n```\\n\\n```\\n1| package main\\n2| \\n3| import \\\"fmt\\\"\\n4| \\n5| \\treturn func() string {\\n6| \\treturn func() string {\\n7| \\t\\treturn \\\"Hello world!\\\"\\n8| \\t}\\n9| }\\n10| \\n11| func main() {\\n12| \\tgreet := makeGreeter()\\n13| \\tfmt.Println(greet())\\n14| \\tfmt.Printf(\\\"%T\\\\n\\\", greet)\\n15| }\\n```\\n\\n## Result 6: github.com/vitorbritto/dexter-momentum/go/concepts/10-functions/stack-function/main.go\\n\\nURL: /r/github.com/vitorbritto/dexter-momentum/-/blob/go/concepts/10-functions/stack-function/main.go\\n\\n```\\n6| \\tdebug.PrintStack()\\n7| }\\n8| \\n9| func f2() {\\n10| \\tf3()\\n11| }\\n12| \\n13| func f1() {\\n14| \\tf2()\\n15| }\\n16| func main() {\\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n```\\n\\n```\\n12| \\n13| func f1() {\\n14| \\tf2()\\n15| }\\n16| \\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // main.f3()\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // main.f1()\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n13| func f1() {\\n14| \\tf2()\\n15| }\\n16| \\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // main.f1()\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n14| \\tf2()\\n15| }\\n16| \\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // main.f2()\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // main.f1()\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n15| }\\n16| \\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // main.f1()\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n16| \\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // main.f1()\\n27| // main.f1()\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n17| func main() {\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n28| // \\t/Users/jason/go/src/stack-function/main.go:11 +0x11\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n```\\n18| \\tf1()\\n19| }\\n20| \\n21| // Output:\\n22| // goroutine 1 [running]:\\n23| // main.f3()\\n24| // \\t/Users/jason/go/src/stack-function/main.go:3 +0x11\\n25| // main.f2()\\n26| // \\t/Users/jason/go/src/stack-function/main.go:7 +0x11\\n27| // main.f1()\\n28| // main.main()\\n29| // main.main()\\n30| // \\t/Users/jason/go/src/stack-function/main.go:15 +0x11\\n31| \\n```\\n\\n\",\"tool_call_id\":\"call_55e4aed5dc874cca8e0a17fb\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -791,607 +485,677 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Found"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"30"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"+"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" matches"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repositories"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Here"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"'s"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" a"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" summary"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" of"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" top"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":\n\n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" #"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Repository"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" File"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Sn"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ippet"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"---"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Found"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-----------"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"30"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"+"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"------"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" matches"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"---------"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repos"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Here"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" are"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" some"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" highlights"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" #"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Repository"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" File"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Description"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"---"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"01"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-----------"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_before"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-f"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"------"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"unc"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-------------"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Basic"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" greeting"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" function"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" called"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"01"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_before"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-f"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"unc"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Simple"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" greeting"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" function"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" called"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"02"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Function"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" assigned"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" variable"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"3"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"02"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Function"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" assigned"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" variable"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"3"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"03"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_sh"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ows"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-type"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Function"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" type"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" printing"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" in"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"03"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_sh"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ows"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"4"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-type"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Function"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" type"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" printing"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"04"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_an"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"other"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-way"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"make"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Gre"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"04"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"eter"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_an"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"other"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" returning"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-way"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" closure"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" called"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Function"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" returning"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" closure"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"oes"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Ele"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ven"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/G"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"05"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"05"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_an"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_an"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"other"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"other"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-way"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-way"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-expression"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_sh"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_sh"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ows"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ows"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-type"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-type"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Same"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Closure"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" pattern"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" type"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" type"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" inspection"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" inspection"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"|"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"web"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"v"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ta"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"itor"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"o"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"brit"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"520"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"to"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/g"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/d"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"olang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"exter"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_project"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-m"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"omentum"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"..."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"panic"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/con"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"_re"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"cepts"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"cover"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"10"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"01"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-functions"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"stack"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-function"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Panic"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/re"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"cover"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" demonstration"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Call"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" stack"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" demonstration"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" are"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" predominantly"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"debug"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".Print"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" training"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Stack"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/tutorial"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"()`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repositories"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" |\n\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Most"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" To"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" narrow"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" are"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" results"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" specific"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" training"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" projects"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/tutorial"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repos"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" you"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" could"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" You"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" refine"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" can"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" narrow"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" query"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" search"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" e"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".g"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" more"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" specific"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" queries"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" like"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" lang"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"func"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" lang"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repo"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"go"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"github"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repo"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".com"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/s"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"specific"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ome"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"-org"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"org"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"/s"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"om"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" target"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"ere"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" particular"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"po"},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" repositories"},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-699e2ea6142ffea5485300c3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":3290,"completion_tokens":301,"total_tokens":11335}} + data: {"id":"chatcmpl-4ac950d7e5e9da5a749757be","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":3117,"completion_tokens":336,"total_tokens":11133}} data: [DONE] @@ -1400,4 +1164,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 3.703162291s + duration: 5.000312125s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/update_a_file.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/update_a_file.yaml index 7c75545c51762a7acd9b2dd0441f6305ee1fe65d..37f43a9085eb7fb7539afeda50038f8e1d2756f2 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/update_a_file.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/update_a_file.yaml @@ -2,6 +2,108 @@ version: 2 interactions: - id: 0 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 33661 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Let"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" find"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" first"},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_2ff8198a89ba48769725fd11","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"pattern\":\"**/main.go\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-82e13d9ff948d6ae1d864e96","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":704,"completion_tokens":25,"total_tokens":7769}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.548065875s +- id: 1 + request: + proto: HTTP/1.1 + proto_major: 1 + proto_minor: 1 + content_length: 34027 + host: "" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_2ff8198a89ba48769725fd11\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_2ff8198a89ba48769725fd11\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + headers: + Accept: + - application/json + Content-Type: + - application/json + User-Agent: + - Charm-Crush/devel (https://charm.land/crush) + url: https://hyper.charm.land/v1/chat/completions + method: POST + response: + proto: HTTP/2.0 + proto_major: 2 + proto_minor: 0 + content_length: -1 + uncompressed: true + body: |+ + data: {"id":"chatcmpl-1d523ed3c3963df51fbe016d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-1d523ed3c3963df51fbe016d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_f59e4cd2d4504618ac2b1ce0","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-1d523ed3c3963df51fbe016d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\"}"}}]},"finish_reason":null}]} + + data: {"id":"chatcmpl-1d523ed3c3963df51fbe016d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":100,"completion_tokens":29,"total_tokens":7809}} + + data: [DONE] + + headers: + Content-Type: + - text/event-stream + status: 200 OK + code: 200 + duration: 3.264337375s +- id: 2 request: proto: HTTP/1.1 proto_major: 1 @@ -25,427 +127,605 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"We"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" need"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" generate"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" based"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" on"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" given"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" message"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" The"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" in"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Generate"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ≤"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" following"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":\""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" then"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" The"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"update"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"update"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" by"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" changing"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" maybe"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Update"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" say"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" summary"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Update"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" same"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" language"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"English"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")."},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Must"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Update"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Possible"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Update"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Update"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"\""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" length"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Update"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" +"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" -\u003e"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" total"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" So"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" OK"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".\n\n"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Return"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" that"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" only"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Update"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" inc"},"finish_reason":null}]} - data: {"id":"chatcmpl-5f2f64db8bf652336ddf40f1","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":5,"completion_tokens":167,"total_tokens":380}} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quote"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?)"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 518.110959ms -- id: 1 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 33661 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Actually"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Let"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" we"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" me"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" probably"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" find"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" shouldn't"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" and"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" include"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" read"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"?"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" could"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" first"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Update"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_c1041a1ca9164f49b628989d","type":"function","index":0,"function":{"name":"glob","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"pattern\":\"**/main.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-edee1f120be199268699c7c2","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":64,"completion_tokens":25,"total_tokens":7769}} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: [DONE] + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} - headers: - Content-Type: - - text/event-stream - status: 200 OK - code: 200 - duration: 5.706396667s -- id: 2 - request: - proto: HTTP/1.1 - proto_major: 1 - proto_minor: 1 - content_length: 34027 - host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_c1041a1ca9164f49b628989d\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_c1041a1ca9164f49b628989d\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" - headers: - Accept: - - application/json - Content-Type: - - application/json - User-Agent: - - Charm-Crush/devel (https://charm.land/crush) - url: https://hyper.charm.land/v1/chat/completions - method: POST - response: - proto: HTTP/2.0 - proto_major: 2 - proto_minor: 0 - content_length: -1 - uncompressed: true - body: |+ - data: {"id":"chatcmpl-cc5d6533243ffa215f301479","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Update"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" U"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" etc"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" let's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" count"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Update"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"6"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"7"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"14"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"15"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"2"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"17"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} - data: {"id":"chatcmpl-cc5d6533243ffa215f301479","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_f670151d1ac842759ca9cc20","type":"function","index":0,"function":{"name":"view","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"18"},"finish_reason":null}]} - data: {"id":"chatcmpl-cc5d6533243ffa215f301479","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-cc5d6533243ffa215f301479","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":36,"completion_tokens":29,"total_tokens":7809}} + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" print"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"23"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"24"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"29"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"30"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" from"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"4"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"34"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" space"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"35"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ("},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"5"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":")="},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"40"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" within"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" colon"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Update"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" main"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" print"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" hello"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" crush"},"finish_reason":null}]} + + data: {"id":"chatcmpl-6933cce4991d9cfc91af725c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":37,"completion_tokens":307,"total_tokens":520}} data: [DONE] @@ -454,7 +734,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 4.151100375s + duration: 645.48275ms - id: 3 request: proto: HTTP/1.1 @@ -462,7 +742,7 @@ interactions: proto_minor: 1 content_length: 34493 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_c1041a1ca9164f49b628989d\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_c1041a1ca9164f49b628989d\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_f670151d1ac842759ca9cc20\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_f670151d1ac842759ca9cc20\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_2ff8198a89ba48769725fd11\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_2ff8198a89ba48769725fd11\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_f59e4cd2d4504618ac2b1ce0\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_f59e4cd2d4504618ac2b1ce0\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -479,13 +759,13 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-1617f1b7ded9cf66036e5397","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-c53ea64c0a8862ccac20675d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-1617f1b7ded9cf66036e5397","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_58db65908ca64b7ebc59e94d","type":"function","index":0,"function":{"name":"edit","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-c53ea64c0a8862ccac20675d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_abfa8c8f93ac4cb9a287f343","type":"function","index":0,"function":{"name":"edit","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-1617f1b7ded9cf66036e5397","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"new_string\":\"\\tfmt.Println(\\\"Hello from crush!\\\")\",\"old_string\":\"\\tfmt.Println(\\\"Hello, World!\\\")\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-c53ea64c0a8862ccac20675d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"new_string\":\"\\tfmt.Println(\\\"Hello from crush!\\\")\",\"old_string\":\"\\tfmt.Println(\\\"Hello, World!\\\")\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-1617f1b7ded9cf66036e5397","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":57,"completion_tokens":55,"total_tokens":7920}} + data: {"id":"chatcmpl-c53ea64c0a8862ccac20675d","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":121,"completion_tokens":55,"total_tokens":7920}} data: [DONE] @@ -494,7 +774,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 5.2384995s + duration: 3.901191333s - id: 4 request: proto: HTTP/1.1 @@ -502,7 +782,7 @@ interactions: proto_minor: 1 content_length: 35034 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_c1041a1ca9164f49b628989d\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_c1041a1ca9164f49b628989d\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_f670151d1ac842759ca9cc20\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_f670151d1ac842759ca9cc20\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_58db65908ca64b7ebc59e94d\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\",\\\"new_string\\\":\\\"\\\\tfmt.Println(\\\\\\\"Hello from crush!\\\\\\\")\\\",\\\"old_string\\\":\\\"\\\\tfmt.Println(\\\\\\\"Hello, World!\\\\\\\")\\\"}\",\"name\":\"edit\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\nContent replaced in file: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\n\\n\",\"tool_call_id\":\"call_58db65908ca64b7ebc59e94d\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"update the main.go file by changing the print to say hello from crush\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_2ff8198a89ba48769725fd11\",\"function\":{\"arguments\":\"{\\\"pattern\\\":\\\"**/main.go\\\"}\",\"name\":\"glob\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Let me find and read the main.go file first.\"},{\"content\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\",\"tool_call_id\":\"call_2ff8198a89ba48769725fd11\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_f59e4cd2d4504618ac2b1ce0\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\"}\",\"name\":\"view\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\n 1|package main\\n 2|\\n 3|import \\\"fmt\\\"\\n 4|\\n 5|func main() {\\n 6|\\tfmt.Println(\\\"Hello, World!\\\")\\n 7|}\\n\\n\",\"tool_call_id\":\"call_f59e4cd2d4504618ac2b1ce0\",\"role\":\"tool\"},{\"tool_calls\":[{\"id\":\"call_abfa8c8f93ac4cb9a287f343\",\"function\":{\"arguments\":\"{\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\\",\\\"new_string\\\":\\\"\\\\tfmt.Println(\\\\\\\"Hello from crush!\\\\\\\")\\\",\\\"old_string\\\":\\\"\\\\tfmt.Println(\\\\\\\"Hello, World!\\\\\\\")\\\"}\",\"name\":\"edit\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"\"},{\"content\":\"\\nContent replaced in file: /tmp/crush-test/TestCoderAgent/glm-5.1/update_a_file/main.go\\n\\n\",\"tool_call_id\":\"call_abfa8c8f93ac4cb9a287f343\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -519,37 +799,41 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Done"},"finish_reason":null}]} + + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Updated"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" Updated"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"main"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".go"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"6"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"6"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" print"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" print"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" from"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" crush"},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" crush"},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"!\""},"finish_reason":null}]} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"!\""},"finish_reason":null}]} - data: {"id":"chatcmpl-6422e5c6cedd8450a094c41a","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":83,"completion_tokens":16,"total_tokens":7971}} + data: {"id":"chatcmpl-10ce50181e34eb27ebcf4d67","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":147,"completion_tokens":18,"total_tokens":7973}} data: [DONE] @@ -558,4 +842,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.301461167s + duration: 3.401118125s diff --git a/internal/agent/testdata/TestCoderAgent/glm-5.1/write_tool.yaml b/internal/agent/testdata/TestCoderAgent/glm-5.1/write_tool.yaml index 131b1cc979a498532bd3605764b2039c3e2fcc68..4793d14a5491e9af480be0418100f1570a3430ef 100644 --- a/internal/agent/testdata/TestCoderAgent/glm-5.1/write_tool.yaml +++ b/internal/agent/testdata/TestCoderAgent/glm-5.1/write_tool.yaml @@ -25,205 +25,209 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"The"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"User"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" user"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" wants"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" concise"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" title"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" for"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" the"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" given"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Need"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" write"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" new"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" called"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" or"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" col"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '{\""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"name"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"test"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\","},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"use"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"version"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" write"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" to"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" create"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" new"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"0"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" called"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"0"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"}"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" '{\""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" should"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"name"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" be"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" short"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"test"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \u003c="},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\","},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"50"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"version"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" one"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" line"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"1"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":","},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" no"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"0"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"/"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"0"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"col"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"}"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ons"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"'"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Summ"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ar"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"ize"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Perhaps"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Create"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" JSON"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" content"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" settings"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Title"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Count"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" characters"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":":"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" \""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"Create"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" config"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":".json"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" JSON"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" data"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" settings"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" That's"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"\""},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" ="},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" "},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"36"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" chars"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" No"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" quotes"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":" Good"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Create"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" config"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":"Create"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".json"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" config"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":".json"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" test"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" with"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" settings"},"finish_reason":null}]} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" test"},"finish_reason":null}]} - data: {"id":"chatcmpl-5de54c80d1a8f3193a84f17c","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":3,"completion_tokens":107,"total_tokens":334}} + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" JSON"},"finish_reason":null}]} + + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{"content":" data"},"finish_reason":null}]} + + data: {"id":"chatcmpl-a5acf5c18b9ab869cdebd547","object":"chat.completion.chunk","created":0,"model":"gpt-oss-120b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":51,"completion_tokens":109,"total_tokens":336}} data: [DONE] @@ -232,7 +236,7 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 400.28525ms + duration: 599.85525ms - id: 1 request: proto: HTTP/1.1 @@ -240,7 +244,7 @@ interactions: proto_minor: 1 content_length: 33698 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use write to create a new file called config.json with content '{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}'\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use write to create a new file called config.json with content '{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}'\",\"role\":\"user\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -257,27 +261,13 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"Simple"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" task"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" -"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" write"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" a"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":" file"},"finish_reason":null}]} - - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"reasoning_content":"."},"finish_reason":null}]} + data: {"id":"chatcmpl-45c1b71f4a24b361d053f9ce","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_ceadb33c267645ef824dbd83","type":"function","index":0,"function":{"name":"write","arguments":""}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-45c1b71f4a24b361d053f9ce","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"id":"call_9ba00bcb7fd94be29ce795ed","type":"function","index":0,"function":{"name":"write","arguments":""}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"content\":\"{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}\",\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\"}"}}]},"finish_reason":null}]} + data: {"id":"chatcmpl-45c1b71f4a24b361d053f9ce","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"{\"content\":\"{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}\",\"file_path\":\"/tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\"}"}}]},"finish_reason":null}]} - data: {"id":"chatcmpl-a968f8262469502ef16a0a60","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":14,"completion_tokens":56,"total_tokens":7814}} + data: {"id":"chatcmpl-45c1b71f4a24b361d053f9ce","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}],"usage":{"prompt_tokens":718,"completion_tokens":49,"total_tokens":7807}} data: [DONE] @@ -286,15 +276,15 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 6.486736834s + duration: 3.932976875s - id: 2 request: proto: HTTP/1.1 proto_major: 1 proto_minor: 1 - content_length: 34227 + content_length: 34177 host: "" - body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use write to create a new file called config.json with content '{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}'\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_ceadb33c267645ef824dbd83\",\"function\":{\"arguments\":\"{\\\"content\\\":\\\"{\\\\\\\"name\\\\\\\": \\\\\\\"test\\\\\\\", \\\\\\\"version\\\\\\\": \\\\\\\"1.0.0\\\\\\\"}\\\",\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\\\"}\",\"name\":\"write\"},\"type\":\"function\"}],\"role\":\"assistant\",\"reasoning_content\":\"Simple task - write a file.\"},{\"content\":\"\\nFile successfully written: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\\n\",\"tool_call_id\":\"call_ceadb33c267645ef824dbd83\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" + body: "{\"messages\":[{\"content\":\"You are Crush, a powerful AI Assistant that runs in the CLI.\\n\\n\\nThese rules override everything else. Follow them strictly:\\n\\n1. **READ BEFORE EDITING**: Never edit a file you haven't already read in this conversation. Once read, you don't need to re-read unless it changed. Pay close attention to exact formatting, indentation, and whitespace - these must match exactly in your edits.\\n2. **BE AUTONOMOUS**: Don't ask questions - search, read, think, decide, act. Break complex tasks into steps and complete them all. Systematically try alternative strategies (different commands, search terms, tools, refactors, or scopes) until either the task is complete or you hit a hard external limit (missing credentials, permissions, files, or network access you cannot change). Only stop for actual blocking errors, not perceived difficulty.\\n3. **TEST AFTER CHANGES**: Run tests immediately after each modification.\\n4. **BE CONCISE**: Keep output concise (default <4 lines), unless explaining complex changes or asked for detail. Conciseness applies to output only, not to thoroughness of work.\\n5. **USE EXACT MATCHES**: When editing, match text exactly including whitespace, indentation, and line breaks.\\n6. **NEVER COMMIT**: Unless user explicitly says \\\"commit\\\". When committing, follow the `` format from the bash tool description exactly, including any configured attribution lines.\\n7. **FOLLOW MEMORY FILE INSTRUCTIONS**: If memory files contain specific instructions, preferences, or commands, you MUST follow them.\\n8. **NEVER ADD COMMENTS**: Only add comments if the user asked you to do so. Focus on *why* not *what*. NEVER communicate with the user through code comments.\\n9. **SECURITY FIRST**: Only assist with defensive security tasks. Refuse to create, modify, or improve code that may be used maliciously.\\n10. **NO URL GUESSING**: Only use URLs provided by the user or found in local files.\\n11. **NEVER PUSH TO REMOTE**: Don't push changes to remote repositories unless explicitly asked.\\n12. **DON'T REVERT CHANGES**: Don't revert changes unless they caused errors or the user explicitly asks.\\n13. **TOOL CONSTRAINTS**: Only use documented tools. Never attempt 'apply_patch' or 'apply_diff' - they don't exist. Use 'edit' or 'multiedit' instead.\\n14. **LOAD MATCHING SKILLS**: If any entry in `` matches the current task, you MUST call `view` on its `` before taking any other action for that task. The `` is only a trigger — the actual procedure, scripts, and references live in SKILL.md. Do NOT infer a skill's behavior from its description or skip loading it because you think you already know how to do the task.\\n\\n\\n\\nKeep responses minimal:\\n- ALWAYS think and respond in the same spoken language the prompt was written in.\\n- Under 4 lines of text (tool use doesn't count)\\n- Conciseness is about **text only**: always fully implement the requested feature, tests, and wiring even if that requires many tool calls.\\n- No preamble (\\\"Here's...\\\", \\\"I'll...\\\")\\n- No postamble (\\\"Let me know...\\\", \\\"Hope this helps...\\\")\\n- One-word answers when possible\\n- No emojis ever\\n- No explanations unless user asks\\n- Never send acknowledgement-only responses; after receiving new context or instructions, immediately continue the task or state the concrete next action you will take.\\n- Use rich Markdown formatting (headings, bullet lists, tables, code fences) for any multi-sentence or explanatory answer; only use plain unformatted text if the user explicitly asks.\\n\\nExamples:\\nuser: what is 2+2?\\nassistant: 4\\n\\nuser: list files in src/\\nassistant: [uses ls tool]\\nfoo.c, bar.c, baz.c\\n\\nuser: which file has the foo implementation?\\nassistant: src/foo.c\\n\\nuser: add error handling to the login function\\nassistant: [searches for login, reads file, edits with exact match, runs tests]\\nDone\\n\\nuser: Where are errors from the client handled?\\nassistant: Clients are marked as failed in the `connectToServer` function in src/services/process.go:712.\\n\\n\\n\\nWhen referencing specific functions or code locations, use the pattern `file_path:line_number` to help users navigate:\\n- Example: \\\"The error is handled in src/main.go:45\\\"\\n- Example: \\\"See the implementation in pkg/utils/helper.go:123-145\\\"\\n\\n\\n\\nFor every task, follow this sequence internally (don't narrate it):\\n\\n**Before acting**:\\n- Search codebase for relevant files\\n- Read files to understand current state\\n- Check memory for stored commands\\n- Identify what needs to change\\n- Use `git log` and `git blame` for additional context when needed\\n\\n**While acting**:\\n- Read entire file before editing it\\n- Before editing: verify exact whitespace and indentation from View output\\n- Use exact text for find/replace (include whitespace)\\n- Make one logical change at a time\\n- After each change: run tests\\n- If tests fail: fix immediately\\n- If edit fails: read more context, don't guess - the text must match exactly\\n- Keep going until query is completely resolved before yielding to user\\n- For longer tasks, send brief progress updates (under 10 words) BUT IMMEDIATELY CONTINUE WORKING - progress updates are not stopping points\\n\\n**Before finishing**:\\n- Verify ENTIRE query is resolved (not just first step)\\n- All described next steps must be completed\\n- Cross-check the original prompt and your own mental checklist; if any feasible part remains undone, continue working instead of responding.\\n- Run lint/typecheck if in memory\\n- Verify all changes work\\n- Keep response under 4 lines\\n\\n**Key behaviors**:\\n- Use find_references before changing shared code\\n- Follow existing patterns (check similar files)\\n- If stuck, try different approach (don't repeat failures)\\n- Make decisions yourself (search first, don't ask)\\n- Fix problems at root cause, not surface-level patches\\n- Don't fix unrelated bugs or broken tests (mention them in final message if relevant)\\n\\n\\n\\n**Make decisions autonomously** - don't ask when you can:\\n- Search to find the answer\\n- Read files to see patterns\\n- Check similar code\\n- Infer from context\\n- Try most likely approach\\n- When requirements are underspecified but not obviously dangerous, make the most reasonable assumptions based on project patterns and memory files, briefly state them if needed, and proceed instead of waiting for clarification.\\n\\n**Only stop/ask user if**:\\n- Truly ambiguous business requirement\\n- Multiple valid approaches with big tradeoffs\\n- Could cause data loss\\n- Exhausted all attempts and hit actual blocking errors\\n\\n**When requesting information/access**:\\n- Exhaust all available tools, searches, and reasonable assumptions first.\\n- Never say \\\"Need more info\\\" without detail.\\n- In the same message, list each missing item, why it is required, acceptable substitutes, and what you already attempted.\\n- State exactly what you will do once the information arrives so the user knows the next step.\\n\\nWhen you must stop, first finish all unblocked parts of the request, then clearly report: (a) what you tried, (b) exactly why you are blocked, and (c) the minimal external action required. Don't stop just because one path failed—exhaust multiple plausible approaches first.\\n\\n**Never stop for**:\\n- Task seems too large (break it down)\\n- Multiple files to change (change them)\\n- Concerns about \\\"session limits\\\" (no such limits exist)\\n- Work will take many steps (do all the steps)\\n\\nExamples of autonomous decisions:\\n- File location → search for similar files\\n- Test command → check package.json/memory\\n- Code style → read existing code\\n- Library choice → check what's used\\n- Naming → follow existing names\\n\\n\\n\\n**Available edit tools:**\\n- `edit` - Single find/replace in a file\\n- `multiedit` - Multiple find/replace operations in one file\\n- `write` - Create/overwrite entire file\\n\\nNever use `apply_patch` or similar - those tools don't exist.\\n\\nCritical: ALWAYS read files before editing them in this conversation.\\n\\nWhen using edit tools:\\n1. Read the file first - note the EXACT indentation (spaces vs tabs, count)\\n2. Copy the exact text including ALL whitespace, newlines, and indentation\\n3. Include 3-5 lines of context before and after the target\\n4. Verify your old_string would appear exactly once in the file\\n5. If uncertain about whitespace, include more surrounding context\\n6. Verify edit succeeded\\n7. Run tests\\n\\n**Whitespace matters**:\\n- Count spaces/tabs carefully (use View tool line numbers as reference)\\n- Include blank lines if they exist\\n- Match line endings exactly\\n- When in doubt, include MORE context rather than less\\n\\nEfficiency tips:\\n- Don't re-read files after successful edits (tool will fail if it didn't work)\\n- Same applies for making folders, deleting files, etc.\\n\\nCommon mistakes to avoid:\\n- Editing without reading first\\n- Approximate text matches\\n- Wrong indentation (spaces vs tabs, wrong count)\\n- Missing or extra blank lines\\n- Not enough context (text appears multiple times)\\n- Trimming whitespace that exists in the original\\n- Not testing after changes\\n\\n\\n\\nThe Edit tool is extremely literal. \\\"Close enough\\\" will fail.\\n\\n**Before every edit**:\\n1. View the file and locate the exact lines to change\\n2. Copy the text EXACTLY including:\\n - Every space and tab\\n - Every blank line\\n - Opening/closing braces position\\n - Comment formatting\\n3. Include enough surrounding lines (3-5) to make it unique\\n4. Double-check indentation level matches\\n\\n**Common failures**:\\n- `func foo() {` vs `func foo(){` (space before brace)\\n- Tab vs 4 spaces vs 2 spaces\\n- Missing blank line before/after\\n- `// comment` vs `//comment` (space after //)\\n- Different number of spaces in indentation\\n\\n**If edit fails**:\\n- View the file again at the specific location\\n- Copy even more context\\n- Check for tabs vs spaces\\n- Verify line endings\\n- Try including the entire function/block if needed\\n- Never retry with guessed changes - get the exact text first\\n\\n\\n\\nEnsure every task is implemented completely, not partially or sketched.\\n\\n1. **Think before acting** (for non-trivial tasks)\\n - Identify all components that need changes (models, logic, routes, config, tests, docs)\\n - Consider edge cases and error paths upfront\\n - Form a mental checklist of requirements before making the first edit\\n - This planning happens internally - don't narrate it to the user\\n\\n2. **Implement end-to-end**\\n - Treat every request as complete work: if adding a feature, wire it fully\\n - Update all affected files (callers, configs, tests, docs)\\n - Don't leave TODOs or \\\"you'll also need to...\\\" - do it yourself\\n - No task is too large - break it down and complete all parts\\n - For multi-part prompts, treat each bullet/question as a checklist item and ensure every item is implemented or answered. Partial completion is not an acceptable final state.\\n\\n3. **Verify before finishing**\\n - Re-read the original request and verify each requirement is met\\n - Check for missing error handling, edge cases, or unwired code\\n - Run tests to confirm the implementation works\\n - Only say \\\"Done\\\" when truly done - never stop mid-task\\n\\n\\n\\nWhen errors occur:\\n1. Read complete error message\\n2. Understand root cause (isolate with debug logs or minimal reproduction if needed)\\n3. Try different approach (don't repeat same action)\\n4. Search for similar code that works\\n5. Make targeted fix\\n6. Test to verify\\n7. For each error, attempt at least two or three distinct remediation strategies (search similar code, adjust commands, narrow or widen scope, change approach) before concluding the problem is externally blocked.\\n\\nCommon errors:\\n- Import/Module → check paths, spelling, what exists\\n- Syntax → check brackets, indentation, typos\\n- Tests fail → read test, see what it expects\\n- File not found → use ls, check exact path\\n\\n**Edit tool \\\"old_string not found\\\"**:\\n- View the file again at the target location\\n- Copy the EXACT text including all whitespace\\n- Include more surrounding context (full function if needed)\\n- Check for tabs vs spaces, extra/missing blank lines\\n- Count indentation spaces carefully\\n- Don't retry with approximate matches - get the exact text\\n\\n\\n\\nMemory files store commands, preferences, and codebase info. Update them when you discover:\\n- Build/test/lint commands\\n- Code style preferences\\n- Important codebase patterns\\n- Useful project information\\n\\n\\n\\nBefore writing code:\\n1. Check if library exists (look at imports, package.json)\\n2. Read similar code for patterns\\n3. Match existing style\\n4. Use same libraries/frameworks\\n5. Follow security best practices (never log secrets)\\n6. Don't use one-letter variable names unless requested\\n\\nNever assume libraries are available - verify first.\\n\\n**Ambition vs. precision**:\\n- New projects → be creative and ambitious with implementation\\n- Existing codebases → be surgical and precise, respect surrounding code\\n- Don't change filenames or variables unnecessarily\\n- Don't add formatters/linters/tests to codebases that don't have them\\n\\n\\n\\nAfter significant changes:\\n- Start testing as specific as possible to code changed, then broaden to build confidence\\n- Use self-verification: write unit tests, add output logs, or use debug statements to verify your solutions\\n- Run relevant test suite\\n- If tests fail, fix before continuing\\n- Check memory for test commands\\n- Run lint/typecheck if available (on precise targets when possible)\\n- For formatters: iterate max 3 times to get it right; if still failing, present correct solution and note formatting issue\\n- Suggest adding commands to memory if not found\\n- Don't fix unrelated bugs or test failures (not your responsibility)\\n\\n\\n\\n- Default to using tools (ls, grep, view, agent, tests, web_fetch, etc.) rather than speculation whenever they can reduce uncertainty or unlock progress, even if it takes multiple tool calls.\\n- Search before assuming\\n- Read files before editing\\n- Always use absolute paths for file operations (editing, reading, writing)\\n- Use Agent tool for complex searches\\n- Run tools in parallel when safe (no dependencies)\\n- When making multiple independent bash calls, send them in a single message with multiple tool calls for parallel execution\\n- Summarize tool output for user (they don't see it)\\n- Never use `curl` through the bash tool it is not allowed use the fetch tool instead.\\n- Only use the tools you know exist.\\n\\n\\n**CRITICAL**: The `description` parameter is REQUIRED for all bash tool calls. Always provide it.\\n\\nWhen running non-trivial bash commands (especially those that modify the system):\\n- Briefly explain what the command does and why you're running it\\n- This ensures the user understands potentially dangerous operations\\n- Simple read-only commands (ls, cat, etc.) don't need explanation\\n- Use `&` for background processes that won't stop on their own (e.g., `node server.js &`)\\n- Avoid interactive commands - use non-interactive versions (e.g., `npm init -y` not `npm init`)\\n- Combine related commands to save time (e.g., `git status && git diff HEAD && git log -n 3`)\\n\\n\\n\\n\\nBalance autonomy with user intent:\\n- When asked to do something → do it fully (including ALL follow-ups and \\\"next steps\\\")\\n- Never describe what you'll do next - just do it\\n- When the user provides new information or clarification, incorporate it immediately and keep executing instead of stopping with an acknowledgement.\\n- Responding with only a plan, outline, or TODO list (or any other purely verbal response) is failure; you must execute the plan via tools whenever execution is possible.\\n- When asked how to approach → explain first, don't auto-implement\\n- After completing work → stop, don't explain (unless asked)\\n- Don't surprise user with unexpected actions\\n\\n\\n\\nAdapt verbosity to match the work completed:\\n\\n**Default (under 4 lines)**:\\n- Simple questions or single-file changes\\n- Casual conversation, greetings, acknowledgements\\n- One-word answers when possible\\n\\n**More detail allowed (up to 10-15 lines)**:\\n- Large multi-file changes that need walkthrough\\n- Complex refactoring where rationale adds value\\n- Tasks where understanding the approach is important\\n- When mentioning unrelated bugs/issues found\\n- Suggesting logical next steps user might want\\n- Structure longer answers with Markdown sections and lists, and put all code, commands, and config in fenced code blocks.\\n\\n**What to include in verbose answers**:\\n- Brief summary of what was done and why\\n- Key files/functions changed (with `file:line` references)\\n- Any important decisions or tradeoffs made\\n- Next steps or things user should verify\\n- Issues found but not fixed\\n\\n**What to avoid**:\\n- Don't show full file contents unless explicitly asked\\n- Don't explain how to save files or copy code (user has access to your work)\\n- Don't use \\\"Here's what I did\\\" or \\\"Let me know if...\\\" style preambles/postambles\\n- Keep tone direct and factual, like handing off work to a teammate\\n\\n\\n\\nWorking directory: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool\\nIs directory a git repo: no\\nPlatform: linux\\nToday's date: 1/1/2025\\n\\n\\n\\n\\n\\n\\n \\n crush-hooks\\n Use when the user wants to add, write, debug, or configure a Crush hook — gating or blocking tool calls, approving or rewriting tool input before execution, injecting context into tool results, or troubleshooting hook behavior in crush.json.\\n crush://skills/crush-hooks/SKILL.md\\n builtin\\n \\n \\n jq\\n Use when the user needs to query, filter, reshape, extract, create, or construct JSON data — including API responses, config files, log output, or any structured data — or when helping the user write or debug JSON transformations.\\n crush://skills/jq/SKILL.md\\n builtin\\n \\n\\n\\n\\nThe `` of each skill is a TRIGGER — it tells you *when* a skill applies. It is NOT a specification of what the skill does or how to do it. The procedure, scripts, commands, references, and required flags live only in the SKILL.md body. You do not know what a skill actually does until you have read its SKILL.md.\\n\\nMANDATORY activation flow:\\n1. Scan `` against the current user task.\\n2. If any skill's `` matches, call the View tool with its `` EXACTLY as shown — before any other tool call that performs the task.\\n3. Read the entire SKILL.md and follow its instructions.\\n4. Only then execute the task, using the skill's prescribed commands/tools.\\n\\nDo NOT skip step 2 because you think you already know how to do the task. Do NOT infer a skill's behavior from its name or description. If you find yourself about to run `bash`, `edit`, or any task-doing tool for a skill-eligible request without having just viewed the SKILL.md, stop and load the skill first.\\n\\nBuiltin skills (type=builtin) use virtual `crush://skills/...` location identifiers. The \\\"crush://\\\" prefix is NOT a URL, network address, or MCP resource — it is a special internal identifier the View tool understands natively. Pass the `` verbatim to View.\\n\\nDo not use MCP tools (including read_mcp_resource) to load skills.\\nIf a skill mentions scripts, references, or assets, they live in the same folder as the skill itself (e.g., scripts/, references/, assets/ subdirectories within the skill's folder).\\n\\n\\n\\n\\n\",\"role\":\"system\"},{\"content\":\"This is a reminder that your todo list is currently empty. DO NOT mention this to the user explicitly because they are already aware.\\nIf you are working on tasks that would benefit from a todo list please use the \\\"todos\\\" tool to create one.\\nIf not, please feel free to ignore. Again do not mention this message to the user.\",\"role\":\"user\"},{\"content\":\"use write to create a new file called config.json with content '{\\\"name\\\": \\\"test\\\", \\\"version\\\": \\\"1.0.0\\\"}'\",\"role\":\"user\"},{\"tool_calls\":[{\"id\":\"call_9ba00bcb7fd94be29ce795ed\",\"function\":{\"arguments\":\"{\\\"content\\\":\\\"{\\\\\\\"name\\\\\\\": \\\\\\\"test\\\\\\\", \\\\\\\"version\\\\\\\": \\\\\\\"1.0.0\\\\\\\"}\\\",\\\"file_path\\\":\\\"/tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\\\"}\",\"name\":\"write\"},\"type\":\"function\"}],\"role\":\"assistant\"},{\"content\":\"\\nFile successfully written: /tmp/crush-test/TestCoderAgent/glm-5.1/write_tool/config.json\\n\",\"tool_call_id\":\"call_9ba00bcb7fd94be29ce795ed\",\"role\":\"tool\"}],\"model\":\"glm-5.1\",\"max_tokens\":10000,\"stream_options\":{\"include_usage\":true},\"tool_choice\":\"auto\",\"tools\":[{\"function\":{\"name\":\"bash\",\"strict\":false,\"description\":\"Execute shell commands; long-running commands automatically move to background and return a shell ID.\\n\\n\\nUses mvdan/sh interpreter (Bash-compatible on all platforms including Windows).\\nUse forward slashes for paths: \\\"ls C:/foo/bar\\\" not \\\"ls C:\\\\foo\\\\bar\\\".\\nCommon shell builtins and core utils available on Windows.\\n\\n\\n\\n1. Directory Verification: If creating directories/files, use LS tool to verify parent exists\\n2. Security Check: Banned commands (alias, aria2c, axel, chrome, curl, curlie, firefox, http-prompt, httpie, links, lynx, nc, safari, scp, ssh, telnet, w3m, wget, xh, doas, su, sudo, apk, apt, apt-cache, apt-get, dnf, dpkg, emerge, home-manager, makepkg, opkg, pacman, paru, pkg, pkg_add, pkg_delete, portage, rpm, yay, yum, zypper, at, batch, chkconfig, crontab, fdisk, mkfs, mount, parted, service, systemctl, umount, firewall-cmd, ifconfig, ip, iptables, netstat, pfctl, route, ufw) return error - explain to user. Safe read-only commands execute without prompts\\n3. Command Execution: Execute with proper quoting, capture output\\n4. Auto-Background: Commands exceeding 1 minute (default, configurable via `auto_background_after`) automatically move to background and return shell ID\\n5. Output Processing: Truncate if exceeds 30000 characters\\n6. Return Result: Include errors, metadata with tags\\n\\n\\n\\n- Command required, working_dir optional (defaults to current directory)\\n- IMPORTANT: Use Grep/Glob/Agent tools instead of 'find'/'grep'. Use View/LS tools instead of 'cat'/'head'/'tail'/'ls'\\n- Chain with ';' or '&&', avoid newlines except in quoted strings\\n- Each command runs in independent shell (no state persistence between calls)\\n- Prefer absolute paths over 'cd' (use 'cd' only if user explicitly requests)\\n\\n\\n\\n- Set run_in_background=true to run commands in a separate background shell\\n- Returns a shell ID for managing the background process\\n- Use job_output tool to view current output from background shell\\n- Use job_kill tool to terminate a background shell\\n- IMPORTANT: NEVER use `&` at the end of commands to run in background - use run_in_background parameter instead\\n- Commands that should run in background:\\n * Long-running servers (e.g., `npm start`, `python -m http.server`, `node server.js`)\\n * Watch/monitoring tasks (e.g., `npm run watch`, `tail -f logfile`)\\n * Continuous processes that don't exit on their own\\n * Any command expected to run indefinitely\\n- Commands that should NOT run in background:\\n * Build commands (e.g., `npm run build`, `go build`)\\n * Test suites (e.g., `npm test`, `pytest`)\\n * Git operations\\n * File operations\\n * Short-lived scripts\\n\\n\\n\\nWhen user asks to create git commit:\\n\\n1. Single message with three tool_use blocks (IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - git log (recent commit message style)\\n\\n2. Add relevant untracked files to staging. Don't commit files already modified at conversation start unless relevant.\\n\\n3. Analyze staged changes in tags:\\n - List changed/added files, summarize nature (feature/enhancement/bug fix/refactoring/test/docs)\\n - Brainstorm purpose/motivation, assess project impact, check for sensitive info\\n - Don't use tools beyond git context\\n - Draft concise (1-2 sentences) message focusing on \\\"why\\\" not \\\"what\\\"\\n - Use clear language, accurate reflection (\\\"add\\\"=new feature, \\\"update\\\"=enhancement, \\\"fix\\\"=bug fix)\\n - Avoid generic messages, review draft\\n\\n4. Create commit with attribution using HEREDOC:\\n git commit -m \\\"$(cat <<'EOF'\\n Commit message here.\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n\\n Co-Authored-By: Crush \\n\\n\\n EOF\\n )\\\"\\n\\n5. If pre-commit hook fails, retry ONCE. If fails again, hook preventing commit. If succeeds but files modified, MUST amend.\\n\\n6. Run git status to verify.\\n\\nNotes: Use \\\"git commit -am\\\" when possible, don't stage unrelated files, NEVER update config, don't push, no -i flags, no empty commits, return empty response.\\n\\n\\n\\nUse gh command for ALL GitHub tasks. When user asks to create PR:\\n\\n1. Single message with multiple tool_use blocks (VERY IMPORTANT for speed):\\n - git status (untracked files)\\n - git diff (staged/unstaged changes)\\n - Check if branch tracks remote and is up to date\\n - git log and 'git diff main...HEAD' (full commit history from main divergence)\\n\\n2. Create new branch if needed\\n3. Commit changes if needed\\n4. Push to remote with -u flag if needed\\n\\n5. Analyze changes in tags:\\n - List commits since diverging from main\\n - Summarize nature of changes\\n - Brainstorm purpose/motivation\\n - Assess project impact\\n - Don't use tools beyond git context\\n - Check for sensitive information\\n - Draft concise (1-2 bullet points) PR summary focusing on \\\"why\\\"\\n - Ensure summary reflects ALL changes since main divergence\\n - Clear, concise language\\n - Accurate reflection of changes and purpose\\n - Avoid generic summaries\\n - Review draft\\n\\n6. Create PR with gh pr create using HEREDOC:\\n gh pr create --title \\\"title\\\" --body \\\"$(cat <<'EOF'\\n\\n ## Summary\\n\\n <1-3 bullet points>\\n\\n ## Test plan\\n\\n [Checklist of TODOs...]\\n\\n\\n \U0001F498 Generated with Crush\\n\\n\\n EOF\\n )\\\"\\n\\nImportant:\\n\\n- Return empty response - user sees gh output\\n- Never update git config\\n\\n\\n\\nGood: pytest /foo/bar/tests\\nBad: cd /foo/bar && pytest tests\\n\\n\",\"parameters\":{\"properties\":{\"auto_background_after\":{\"description\":\"Seconds to wait before automatically moving the command to a background job (default: 60)\",\"type\":\"integer\"},\"command\":{\"description\":\"The command to execute\",\"type\":\"string\"},\"description\":{\"description\":\"A brief description of what the command does, try to keep it under 30 characters or so\",\"type\":\"string\"},\"run_in_background\":{\"description\":\"Set to true (boolean) to run this command in the background. Use job_output to read the output later.\",\"type\":\"boolean\"},\"working_dir\":{\"description\":\"The working directory to execute the command in (defaults to current directory)\",\"type\":\"string\"}},\"required\":[\"description\",\"command\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"download\",\"strict\":false,\"description\":\"Download a URL directly to a local file (binary-safe, streaming, max 100MB); overwrites without warning. For reading content into context use fetch.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The local file path where the downloaded content should be saved\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 600)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to download from\",\"type\":\"string\"}},\"required\":[\"url\",\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"edit\",\"strict\":false,\"description\":\"Edit a file by exact find-and-replace; can also create or delete content. For renames/moves use bash. For large edits use write.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"},\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false)\",\"type\":\"boolean\"}},\"required\":[\"file_path\",\"old_string\",\"new_string\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"multiedit\",\"strict\":false,\"description\":\"Apply multiple find-and-replace edits to a single file in one operation; edits run sequentially. Prefer over edit for multiple changes to the same file. Same exact-match rules as edit apply.\",\"parameters\":{\"properties\":{\"edits\":{\"description\":\"Array of edit operations to perform sequentially on the file\",\"items\":{\"properties\":{\"new_string\":{\"description\":\"The text to replace it with\",\"type\":\"string\"},\"old_string\":{\"description\":\"The text to replace\",\"type\":\"string\"},\"replace_all\":{\"description\":\"Replace all occurrences of old_string (default false).\",\"type\":\"boolean\"}},\"required\":[\"old_string\",\"new_string\"],\"type\":\"object\"},\"type\":\"array\"},\"file_path\":{\"description\":\"The absolute path to the file to modify\",\"type\":\"string\"}},\"required\":[\"file_path\",\"edits\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"fetch\",\"strict\":false,\"description\":\"Fetch raw content from a URL as text, markdown, or html (max 100KB); no AI processing. For analysis or extraction use agentic_fetch.\",\"parameters\":{\"properties\":{\"format\":{\"description\":\"The format to return the content in (text, markdown, or html)\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"},\"url\":{\"description\":\"The URL to fetch content from\",\"type\":\"string\"}},\"required\":[\"url\",\"format\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"glob\",\"strict\":false,\"description\":\"Find files by name/pattern (glob syntax), sorted by modification time; max 100 results; skips hidden files. Use grep to search file contents.\",\"parameters\":{\"properties\":{\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The glob pattern to match files against\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"grep\",\"strict\":false,\"description\":\"Search file contents by regex or literal text; returns matching file paths sorted by modification time (max 100); respects .gitignore. Use glob to filter by filename, not contents.\",\"parameters\":{\"properties\":{\"include\":{\"description\":\"File pattern to include in the search (e.g. \\\"*.js\\\", \\\"*.{ts,tsx}\\\")\",\"type\":\"string\"},\"literal_text\":{\"description\":\"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.\",\"type\":\"boolean\"},\"path\":{\"description\":\"The directory to search in. Defaults to the current working directory.\",\"type\":\"string\"},\"pattern\":{\"description\":\"The regex pattern to search for in file contents\",\"type\":\"string\"}},\"required\":[\"pattern\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"ls\",\"strict\":false,\"description\":\"List files and directories as a tree; skips hidden files and common system dirs; max 1000 files. Use glob to find files by pattern, grep to search contents.\",\"parameters\":{\"properties\":{\"depth\":{\"description\":\"The maximum depth to traverse\",\"type\":\"integer\"},\"ignore\":{\"description\":\"List of glob patterns to ignore\",\"items\":{\"type\":\"string\"},\"type\":\"array\"},\"path\":{\"description\":\"The path to the directory to list (defaults to current working directory)\",\"type\":\"string\"}},\"required\":[],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"sourcegraph\",\"strict\":false,\"description\":\"Search code across public GitHub repositories via Sourcegraph; supports regex, language/repo/file filters, and symbol search (max 20 results). Only searches public repos.\",\"parameters\":{\"properties\":{\"context_window\":{\"description\":\"The context around the match to return (default: 10 lines)\",\"type\":\"integer\"},\"count\":{\"description\":\"Optional number of results to return (default: 10, max: 20)\",\"type\":\"integer\"},\"query\":{\"description\":\"The Sourcegraph search query\",\"type\":\"string\"},\"timeout\":{\"description\":\"Optional timeout in seconds (max 120)\",\"type\":\"integer\"}},\"required\":[\"query\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"view\",\"strict\":false,\"description\":\"Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories.\",\"parameters\":{\"properties\":{\"file_path\":{\"description\":\"The path to the file to read\",\"type\":\"string\"},\"limit\":{\"description\":\"The number of lines to read (defaults to 2000)\",\"type\":\"integer\"},\"offset\":{\"description\":\"The line number to start reading from (0-based)\",\"type\":\"integer\"}},\"required\":[\"file_path\"],\"type\":\"object\"}},\"type\":\"function\"},{\"function\":{\"name\":\"write\",\"strict\":false,\"description\":\"Create or overwrite a file with given content; auto-creates parent dirs. Cannot append. Read the file first to avoid conflicts. For surgical changes use edit or multiedit.\",\"parameters\":{\"properties\":{\"content\":{\"description\":\"The content to write to the file\",\"type\":\"string\"},\"file_path\":{\"description\":\"The path to the file to write\",\"type\":\"string\"}},\"required\":[\"file_path\",\"content\"],\"type\":\"object\"}},\"type\":\"function\"}],\"stream\":true}" headers: Accept: - application/json @@ -311,19 +301,19 @@ interactions: content_length: -1 uncompressed: true body: |+ - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Created"},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"Created"},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":" `"},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"config"},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"config"},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".json"},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":".json"},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{"content":"`."},"finish_reason":null}]} - data: {"id":"chatcmpl-07add96c7dd1c61fa3d439d3","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":32,"completion_tokens":7,"total_tokens":7847}} + data: {"id":"chatcmpl-27b68ad28608ebba6c080f72","object":"chat.completion.chunk","created":0,"model":"glm-5.1","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":96,"completion_tokens":7,"total_tokens":7847}} data: [DONE] @@ -332,4 +322,4 @@ interactions: - text/event-stream status: 200 OK code: 200 - duration: 2.346320208s + duration: 2.974104334s diff --git a/internal/agent/tools/job_kill.go b/internal/agent/tools/job_kill.go index 90a44cb131da16a68a4c6cafcd0c6a3f7b2605a9..575496d96714d34147e97b5ca2aefaacd6c68df5 100644 --- a/internal/agent/tools/job_kill.go +++ b/internal/agent/tools/job_kill.go @@ -29,7 +29,7 @@ type JobKillResponseMetadata struct { func NewJobKillTool() fantasy.AgentTool { return fantasy.NewAgentTool( JobKillToolName, - FirstLineDescription(jobKillDescription), + string(jobKillDescription), func(ctx context.Context, params JobKillParams, call fantasy.ToolCall) (fantasy.ToolResponse, error) { if params.ShellID == "" { return fantasy.NewTextErrorResponse("missing shell_id"), nil diff --git a/internal/agent/tools/job_kill.md b/internal/agent/tools/job_kill.md index 96fbbff7f3eeee6265a4c7e401dc603edd47b96e..276a25aa9aa82c201feda5f97659f3580aa9f9ff 100644 --- a/internal/agent/tools/job_kill.md +++ b/internal/agent/tools/job_kill.md @@ -1,4 +1,4 @@ -Terminate a background shell process by ID; shell ID becomes invalid after killing. +Terminate a background shell process. - Provide the shell ID returned from a background bash execution diff --git a/internal/agent/tools/view.go b/internal/agent/tools/view.go index 6880679804a1089fe6e765273bc375e126c817ad..3956a998698233aa194172551b5ab4f69b0d3a30 100644 --- a/internal/agent/tools/view.go +++ b/internal/agent/tools/view.go @@ -53,7 +53,7 @@ type ViewResponseMetadata struct { const ( ViewToolName = "view" - MaxViewSize = 100 * 1024 // 100KB + MaxViewSize = 200 * 1024 // 200KB DefaultReadLimit = 2000 MaxLineLength = 2000 ) diff --git a/internal/agent/tools/view.md b/internal/agent/tools/view.md index 598d90da75b77d183791419c7d34efdea090f7c8..bfeada26efba57b33b771fdaffa1366f570a91eb 100644 --- a/internal/agent/tools/view.md +++ b/internal/agent/tools/view.md @@ -1,4 +1,4 @@ -Read a file by path with line numbers; supports offset and line limit (default 2000, max 100KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories. +Read a file by path with line numbers; supports offset and line limit (default 2000, max 200KB); renders images (PNG, JPEG, GIF, BMP, SVG, WebP); use ls for directories. - Provide file path to read @@ -18,7 +18,7 @@ Read a file by path with line numbers; supports offset and line limit (default 2 -- Max file size: 100KB +- Max file size: 200KB - Default limit: 2000 lines - Lines >2000 chars truncated - Binary files (except images) cannot be displayed diff --git a/internal/app/app.go b/internal/app/app.go index fe31bda6f03d50cd53c3811568858bbfc9316974..a167ca8638c8497a6d6f4260782ba334c6dbe0c3 100644 --- a/internal/app/app.go +++ b/internal/app/app.go @@ -66,7 +66,7 @@ type App struct { serviceEventsWG *sync.WaitGroup eventsCtx context.Context - events chan tea.Msg + events *pubsub.Broker[tea.Msg] tuiWG *sync.WaitGroup // global context and cleanup functions @@ -100,7 +100,7 @@ func New(ctx context.Context, conn *sql.DB, store *config.ConfigStore) (*App, er config: store, - events: make(chan tea.Msg, 100), + events: pubsub.NewBroker[tea.Msg](), serviceEventsWG: &sync.WaitGroup{}, tuiWG: &sync.WaitGroup{}, agentNotifications: pubsub.NewBroker[notify.Notification](), @@ -153,18 +153,15 @@ func (app *App) Store() *config.ConfigStore { return app.config } -// Events returns the events channel for the application. -func (app *App) Events() <-chan tea.Msg { - return app.events +// Events returns a per-caller subscription channel for application events. +// Each caller receives its own channel; all callers receive every event. +func (app *App) Events(ctx context.Context) <-chan pubsub.Event[tea.Msg] { + return app.events.Subscribe(ctx) } -// SendEvent pushes a message into the application's events channel. -// It is non-blocking; the message is dropped if the channel is full. +// SendEvent publishes a message to all event subscribers. func (app *App) SendEvent(msg tea.Msg) { - select { - case app.events <- msg: - default: - } + app.events.Publish(pubsub.UpdatedEvent, msg) } // AgentNotifications returns the broker for agent notification events. @@ -233,7 +230,7 @@ func (app *App) RunNonInteractive(ctx context.Context, output io.Writer, prompt, progress = app.config.Config().Options.Progress == nil || *app.config.Config().Options.Progress if !hideSpinner && stderrTTY { - t := styles.CharmtonePantera() + t := styles.ThemeForProvider(app.config.Config().Models[config.SelectedModelTypeLarge].Provider) // Detect background color to set the appropriate color for the // spinner's 'Generating...' text. Without this, that text would be @@ -486,26 +483,21 @@ func (app *App) setupEvents() { cleanupFunc := func(context.Context) error { cancel() app.serviceEventsWG.Wait() + app.events.Shutdown() return nil } app.cleanupFuncs = append(app.cleanupFuncs, cleanupFunc) } -const subscriberSendTimeout = 2 * time.Second - func setupSubscriber[T any]( ctx context.Context, wg *sync.WaitGroup, name string, subscriber func(context.Context) <-chan pubsub.Event[T], - outputCh chan<- tea.Msg, + broker *pubsub.Broker[tea.Msg], ) { wg.Go(func() { subCh := subscriber(ctx) - sendTimer := time.NewTimer(0) - <-sendTimer.C - defer sendTimer.Stop() - for { select { case event, ok := <-subCh: @@ -513,23 +505,7 @@ func setupSubscriber[T any]( slog.Debug("Subscription channel closed", "name", name) return } - var msg tea.Msg = event - if !sendTimer.Stop() { - select { - case <-sendTimer.C: - default: - } - } - sendTimer.Reset(subscriberSendTimeout) - - select { - case outputCh <- msg: - case <-sendTimer.C: - slog.Debug("Message dropped due to slow consumer", "name", name) - case <-ctx.Done(): - slog.Debug("Subscription cancelled", "name", name) - return - } + broker.Publish(pubsub.UpdatedEvent, tea.Msg(event)) case <-ctx.Done(): slog.Debug("Subscription cancelled", "name", name) return @@ -579,17 +555,18 @@ func (app *App) Subscribe(program *tea.Program) { }) defer app.tuiWG.Done() + events := app.events.Subscribe(tuiCtx) for { select { case <-tuiCtx.Done(): slog.Debug("TUI message handler shutting down") return - case msg, ok := <-app.events: + case ev, ok := <-events: if !ok { slog.Debug("TUI message channel closed") return } - program.Send(msg) + program.Send(ev.Payload) } } } @@ -649,9 +626,9 @@ func (app *App) checkForUpdates(ctx context.Context) { if err != nil || !info.Available() { return } - app.events <- UpdateAvailableMsg{ + app.events.Publish(pubsub.UpdatedEvent, UpdateAvailableMsg{ CurrentVersion: info.Current, LatestVersion: info.Latest, IsDevelopment: info.IsDevelopment(), - } + }) } diff --git a/internal/app/app_test.go b/internal/app/app_test.go index 61b99158f9979d7e21a3c9fe7ad19c74a8111242..e87846ed9744fbfa950867a5e7baae2160518771 100644 --- a/internal/app/app_test.go +++ b/internal/app/app_test.go @@ -2,156 +2,182 @@ package app import ( "context" + "fmt" "sync" "testing" - "testing/synctest" "time" tea "charm.land/bubbletea/v2" "github.com/charmbracelet/crush/internal/pubsub" "github.com/stretchr/testify/require" - "go.uber.org/goleak" ) +// TestSetupSubscriber_NormalFlow verifies that events published to the source +// broker are forwarded to the output broker. func TestSetupSubscriber_NormalFlow(t *testing.T) { - synctest.Test(t, func(t *testing.T) { - f := newSubscriberFixture(t, 10) + t.Parallel() - time.Sleep(10 * time.Millisecond) - synctest.Wait() + ctx, cancel := context.WithCancel(t.Context()) + defer cancel() - f.broker.Publish(pubsub.CreatedEvent, "event1") - f.broker.Publish(pubsub.CreatedEvent, "event2") + src := pubsub.NewBroker[string]() + defer src.Shutdown() + out := pubsub.NewBroker[tea.Msg]() + defer out.Shutdown() - for range 2 { - select { - case <-f.outputCh: - case <-time.After(5 * time.Second): - t.Fatal("Timed out waiting for messages") - } + ch := out.Subscribe(ctx) + + var wg sync.WaitGroup + setupSubscriber(ctx, &wg, "test", src.Subscribe, out) + + // Yield so the subscriber goroutine can call src.Subscribe before we publish. + time.Sleep(10 * time.Millisecond) + + src.Publish(pubsub.CreatedEvent, "hello") + src.Publish(pubsub.CreatedEvent, "world") + + for range 2 { + select { + case <-ch: + case <-time.After(5 * time.Second): + t.Fatal("timed out waiting for forwarded event") } + } - f.cancel() - f.wg.Wait() - }) + cancel() + wg.Wait() } -func TestSetupSubscriber_SlowConsumer(t *testing.T) { - synctest.Test(t, func(t *testing.T) { - f := newSubscriberFixture(t, 0) +// TestSetupSubscriber_ContextCancellation verifies the goroutine exits cleanly +// when the context is cancelled. +func TestSetupSubscriber_ContextCancellation(t *testing.T) { + t.Parallel() - const numEvents = 5 + ctx, cancel := context.WithCancel(t.Context()) - var pubWg sync.WaitGroup - pubWg.Go(func() { - for range numEvents { - f.broker.Publish(pubsub.CreatedEvent, "event") - time.Sleep(10 * time.Millisecond) - synctest.Wait() - } - }) + src := pubsub.NewBroker[string]() + defer src.Shutdown() + out := pubsub.NewBroker[tea.Msg]() + defer out.Shutdown() - time.Sleep(time.Duration(numEvents) * (subscriberSendTimeout + 20*time.Millisecond)) - synctest.Wait() - - received := 0 - for { - select { - case <-f.outputCh: - received++ - default: - pubWg.Wait() - f.cancel() - f.wg.Wait() - require.Less(t, received, numEvents, "Slow consumer should have dropped some messages") - return - } - } - }) + var wg sync.WaitGroup + setupSubscriber(ctx, &wg, "test", src.Subscribe, out) + + src.Publish(pubsub.CreatedEvent, "event") + cancel() + + done := make(chan struct{}) + go func() { wg.Wait(); close(done) }() + + select { + case <-done: + case <-time.After(5 * time.Second): + t.Fatal("setupSubscriber goroutine did not exit after context cancellation") + } } -func TestSetupSubscriber_ContextCancellation(t *testing.T) { - synctest.Test(t, func(t *testing.T) { - f := newSubscriberFixture(t, 10) +// TestEvents_ZeroConsumers verifies that publishing with no subscribers does +// not block or panic. +func TestEvents_ZeroConsumers(t *testing.T) { + t.Parallel() + + broker := pubsub.NewBroker[tea.Msg]() + defer broker.Shutdown() + + require.Equal(t, 0, broker.GetSubscriberCount()) - f.broker.Publish(pubsub.CreatedEvent, "event1") - time.Sleep(100 * time.Millisecond) - synctest.Wait() + // Must not block. + done := make(chan struct{}) + go func() { + broker.Publish(pubsub.UpdatedEvent, tea.Msg("msg1")) + broker.Publish(pubsub.UpdatedEvent, tea.Msg("msg2")) + close(done) + }() - f.cancel() - f.wg.Wait() - }) + select { + case <-done: + case <-time.After(time.Second): + t.Fatal("Publish with zero consumers blocked") + } } -func TestSetupSubscriber_DrainAfterDrop(t *testing.T) { - synctest.Test(t, func(t *testing.T) { - f := newSubscriberFixture(t, 0) +// TestEvents_OneConsumer verifies that a single subscriber receives every event +// exactly once. +func TestEvents_OneConsumer(t *testing.T) { + t.Parallel() - time.Sleep(10 * time.Millisecond) - synctest.Wait() + ctx, cancel := context.WithCancel(t.Context()) + defer cancel() - // First event: nobody reads outputCh so the timer fires (message dropped). - f.broker.Publish(pubsub.CreatedEvent, "event1") - time.Sleep(subscriberSendTimeout + 25*time.Millisecond) - synctest.Wait() + broker := pubsub.NewBroker[tea.Msg]() + defer broker.Shutdown() - // Second event: triggers Stop()==false path; without the fix this deadlocks. - f.broker.Publish(pubsub.CreatedEvent, "event2") + ch := broker.Subscribe(ctx) - // If the timer drain deadlocks, wg.Wait never returns. - done := make(chan struct{}) - go func() { - f.cancel() - f.wg.Wait() - close(done) - }() + const n = 10 + for i := range n { + broker.Publish(pubsub.UpdatedEvent, tea.Msg(i)) + } + for i := range n { select { - case <-done: + case ev := <-ch: + require.Equal(t, tea.Msg(i), ev.Payload) case <-time.After(5 * time.Second): - t.Fatal("setupSubscriber goroutine hung — likely timer drain deadlock") + t.Fatalf("timed out waiting for event %d", i) } - }) + } } -func TestSetupSubscriber_NoTimerLeak(t *testing.T) { - defer goleak.VerifyNone(t) - synctest.Test(t, func(t *testing.T) { - f := newSubscriberFixture(t, 100) +// TestEvents_NConsumers verifies that every subscriber receives every event +// exactly once, regardless of how many concurrent consumers are attached. +func TestEvents_NConsumers(t *testing.T) { + t.Parallel() - for range 100 { - f.broker.Publish(pubsub.CreatedEvent, "event") - time.Sleep(5 * time.Millisecond) - synctest.Wait() - } - - f.cancel() - f.wg.Wait() - }) -} - -type subscriberFixture struct { - broker *pubsub.Broker[string] - wg sync.WaitGroup - outputCh chan tea.Msg - cancel context.CancelFunc + for _, n := range []int{2, 5, 10} { + t.Run(fmt.Sprintf("consumers=%d", n), func(t *testing.T) { + t.Parallel() + testNConsumers(t, n) + }) + } } -func newSubscriberFixture(t *testing.T, bufSize int) *subscriberFixture { +func testNConsumers(t *testing.T, n int) { t.Helper() + ctx, cancel := context.WithCancel(t.Context()) - t.Cleanup(cancel) + defer cancel() + + broker := pubsub.NewBroker[tea.Msg]() + defer broker.Shutdown() - f := &subscriberFixture{ - broker: pubsub.NewBroker[string](), - outputCh: make(chan tea.Msg, bufSize), - cancel: cancel, + // Subscribe all N consumers before publishing. + channels := make([]<-chan pubsub.Event[tea.Msg], n) + for i := range n { + channels[i] = broker.Subscribe(ctx) } - t.Cleanup(f.broker.Shutdown) + require.Equal(t, n, broker.GetSubscriberCount()) - setupSubscriber(ctx, &f.wg, "test", func(ctx context.Context) <-chan pubsub.Event[string] { - return f.broker.Subscribe(ctx) - }, f.outputCh) + const numEvents = 20 + for i := range numEvents { + broker.Publish(pubsub.UpdatedEvent, tea.Msg(i)) + } - return f + // Each consumer must receive all numEvents messages. + var wg sync.WaitGroup + for i, ch := range channels { + wg.Go(func() { + for j := range numEvents { + select { + case ev := <-ch: + require.Equal(t, tea.Msg(j), ev.Payload, + "consumer %d: wrong payload for event %d", i, j) + case <-time.After(5 * time.Second): + t.Errorf("consumer %d: timed out waiting for event %d", i, j) + return + } + } + }) + } + wg.Wait() } diff --git a/internal/backend/events.go b/internal/backend/events.go index a91bad1d5322d1c0ed909b3239e9e97c0eb0c366..6df61f8a742f6672582a3e0db0ba6afbbc2d53af 100644 --- a/internal/backend/events.go +++ b/internal/backend/events.go @@ -8,16 +8,18 @@ import ( mcptools "github.com/charmbracelet/crush/internal/agent/tools/mcp" "github.com/charmbracelet/crush/internal/app" "github.com/charmbracelet/crush/internal/config" + "github.com/charmbracelet/crush/internal/pubsub" ) -// SubscribeEvents returns the event channel for a workspace's app. -func (b *Backend) SubscribeEvents(workspaceID string) (<-chan tea.Msg, error) { +// SubscribeEvents returns a per-caller event channel for a workspace. +// Each caller receives all events; multiple callers do not compete. +func (b *Backend) SubscribeEvents(ctx context.Context, workspaceID string) (<-chan pubsub.Event[tea.Msg], error) { ws, err := b.GetWorkspace(workspaceID) if err != nil { return nil, err } - return ws.Events(), nil + return ws.Events(ctx), nil } // GetLSPStates returns the state of all LSP clients. diff --git a/internal/client/dial_other.go b/internal/client/dial_other.go index f2ba8569ba3326f2df82dc34bbf842eac30918d9..59fa77f4ce793ec9c8d1dc64ae904bb00c9dbefb 100644 --- a/internal/client/dial_other.go +++ b/internal/client/dial_other.go @@ -1,5 +1,4 @@ //go:build !windows -// +build !windows package client diff --git a/internal/cmd/login.go b/internal/cmd/login.go index 9ce9a3e28deb168f7a78b38417e8d93d02ae69ce..1d99468f5027690505f6e8248470ff28871389f4 100644 --- a/internal/cmd/login.go +++ b/internal/cmd/login.go @@ -9,7 +9,6 @@ import ( "charm.land/lipgloss/v2" "github.com/atotto/clipboard" - hyperp "github.com/charmbracelet/crush/internal/agent/hyper" "github.com/charmbracelet/crush/internal/client" "github.com/charmbracelet/crush/internal/config" "github.com/charmbracelet/crush/internal/oauth" @@ -70,9 +69,6 @@ crush login copilot } func loginHyper(c *client.Client, wsID string) error { - if !hyperp.Enabled() { - return fmt.Errorf("hyper not enabled") - } ctx := getLoginContext() resp, err := hyper.InitiateDeviceAuth(ctx) diff --git a/internal/cmd/root_other.go b/internal/cmd/root_other.go index 6d178a07a6e55c85c7fdd4d6a4d98d923aad5a71..de22834b6bdeee5f98d861fc1a5ff86a25d4dd11 100644 --- a/internal/cmd/root_other.go +++ b/internal/cmd/root_other.go @@ -1,5 +1,4 @@ //go:build !windows -// +build !windows package cmd diff --git a/internal/cmd/run.go b/internal/cmd/run.go index 4f73fe5644b9c7ba39ec8d4246c2dc72773469e6..d957720068c47b85038731d7c4ec6eb2dbe9c135 100644 --- a/internal/cmd/run.go +++ b/internal/cmd/run.go @@ -189,7 +189,7 @@ func runNonInteractive( progress = ws.Config.Options.Progress == nil || *ws.Config.Options.Progress if !hideSpinner && stderrTTY { - t := styles.CharmtonePantera() + t := styles.ThemeForProvider(ws.Config.Models[config.SelectedModelTypeLarge].Provider) hasDarkBG := true if stdinTTY && stdoutTTY { diff --git a/internal/cmd/server_other.go b/internal/cmd/server_other.go index 58b05629bf5b85a579ded6379dec53f555fb68e7..25b1281b166b1cc6ce9ea3ad565ef3256c1e1c92 100644 --- a/internal/cmd/server_other.go +++ b/internal/cmd/server_other.go @@ -1,5 +1,4 @@ //go:build !windows -// +build !windows package cmd diff --git a/internal/cmd/session.go b/internal/cmd/session.go index d673ad07c3d11191506498ff4f97410eb1177596..31765cd4aa30cbcf849376addbd3abac2ede4e16 100644 --- a/internal/cmd/session.go +++ b/internal/cmd/session.go @@ -101,6 +101,7 @@ func init() { type sessionServices struct { sessions session.Service messages message.Service + cfg *config.ConfigStore } func sessionSetup(cmd *cobra.Command) (context.Context, *sessionServices, func(), error) { @@ -127,6 +128,7 @@ func sessionSetup(cmd *cobra.Command) (context.Context, *sessionServices, func() svc := &sessionServices{ sessions: session.NewService(queries, conn), messages: message.NewService(queries), + cfg: cfg, } return ctx, svc, func() { conn.Close() }, nil } @@ -280,7 +282,7 @@ func runSessionShow(cmd *cobra.Command, args []string) error { if sessionShowJSON { return outputSessionJSON(cmd.OutOrStdout(), sess, msgPtrs) } - return outputSessionHuman(ctx, sess, msgPtrs) + return outputSessionHuman(ctx, svc.cfg, sess, msgPtrs) } func runSessionDelete(cmd *cobra.Command, args []string) error { @@ -387,7 +389,7 @@ func runSessionLast(cmd *cobra.Command, _ []string) error { if sessionLastJSON { return outputSessionJSON(cmd.OutOrStdout(), sess, msgPtrs) } - return outputSessionHuman(ctx, sess, msgPtrs) + return outputSessionHuman(ctx, svc.cfg, sess, msgPtrs) } const ( @@ -437,8 +439,12 @@ func outputSessionJSON(w io.Writer, sess session.Session, msgs []*message.Messag return enc.Encode(output) } -func outputSessionHuman(ctx context.Context, sess session.Session, msgs []*message.Message) error { - styles := styles.CharmtonePantera() +func outputSessionHuman(ctx context.Context, cfg *config.ConfigStore, sess session.Session, msgs []*message.Message) error { + var providerID string + if cfg != nil { + providerID = cfg.Config().Models[config.SelectedModelTypeLarge].Provider + } + styles := styles.ThemeForProvider(providerID) toolResults := chat.BuildToolResultMap(msgs) width := sessionOutputWidth diff --git a/internal/config/config.go b/internal/config/config.go index 1b8156c68f30b7eddf75d78c414f03662d94f5f2..33251615d44252aaeb2b5db58577759b5dfdff51 100644 --- a/internal/config/config.go +++ b/internal/config/config.go @@ -9,7 +9,6 @@ import ( "maps" "net/http" "net/url" - "regexp" "slices" "strings" "time" @@ -371,7 +370,9 @@ func (t ToolGrep) GetTimeout() time.Duration { } // HookConfig defines a user-configured shell command that fires on a hook -// event (e.g. PreToolUse). +// event (e.g. PreToolUse). This is a pure-data struct: matcher compilation +// is owned by hooks.Runner so a JSON round-trip, merge, or reload can't +// silently drop compiled state. type HookConfig struct { // Regex pattern tested against the tool name. Empty means match all. Matcher string `json:"matcher,omitempty" jsonschema:"description=Regex pattern tested against the tool name. Empty means match all tools."` @@ -379,15 +380,6 @@ type HookConfig struct { Command string `json:"command" jsonschema:"required,description=Shell command to execute when the hook fires"` // Timeout in seconds. Default 30. Timeout int `json:"timeout,omitempty" jsonschema:"description=Timeout in seconds for the hook command,default=30"` - - // Compiled matcher regex. Not serialized. - matcherRegex *regexp.Regexp -} - -// MatcherRegex returns the compiled matcher regex, or nil if no matcher is -// set. -func (h *HookConfig) MatcherRegex() *regexp.Regexp { - return h.matcherRegex } // TimeoutDuration returns the hook timeout as a time.Duration, defaulting diff --git a/internal/config/load.go b/internal/config/load.go index d9f6807e290d97bfb7418b1def2ed471514f8335..967af2de8cd6bffd6a8bd08e77b562998a8d1913 100644 --- a/internal/config/load.go +++ b/internal/config/load.go @@ -905,8 +905,11 @@ func normalizeHookEvent(name string) string { } } -// ValidateHooks normalizes event names and compiles matcher regexes for all -// configured hooks. Returns an error if any regex is invalid. +// ValidateHooks normalizes event names and checks that every configured +// hook has a command and a syntactically valid matcher regex. Matcher +// compilation used for matching is owned by hooks.Runner; this function +// only validates up front so the user sees config errors at load time +// rather than on the first tool call. func (c *Config) ValidateHooks() error { // Normalize event name keys. for event, eventHooks := range c.Hooks { @@ -918,17 +921,15 @@ func (c *Config) ValidateHooks() error { } for event, eventHooks := range c.Hooks { - for i := range eventHooks { - h := &c.Hooks[event][i] + for i, h := range eventHooks { if h.Command == "" { return fmt.Errorf("hook %s[%d]: command is required", event, i) } - if h.Matcher != "" { - re, err := regexp.Compile(h.Matcher) - if err != nil { - return fmt.Errorf("hook %s[%d]: invalid matcher regex %q: %w", event, i, h.Matcher, err) - } - h.matcherRegex = re + if h.Matcher == "" { + continue + } + if _, err := regexp.Compile(h.Matcher); err != nil { + return fmt.Errorf("hook %s[%d]: invalid matcher regex %q: %w", event, i, h.Matcher, err) } } } diff --git a/internal/config/provider.go b/internal/config/provider.go index 645f863628edb8335782fa6c4424ee36f1b10f0e..dffd4dd854f029bef250a1dde9ca557ace4d33c2 100644 --- a/internal/config/provider.go +++ b/internal/config/provider.go @@ -91,9 +91,6 @@ func UpdateProviders(pathOrURL string) error { // UpdateHyper updates the Hyper provider information from a specified URL. func UpdateHyper(pathOrURL string) error { - if !hyper.Enabled() { - return fmt.Errorf("hyper not enabled") - } var provider catwalk.Provider pathOrURL = cmp.Or(pathOrURL, hyper.BaseURL()) @@ -169,7 +166,7 @@ func Providers(cfg *Config) ([]catwalk.Provider, error) { }) wg.Go(func() { - if customProvidersOnly || !hyper.Enabled() { + if customProvidersOnly { return } path := cachePathFor("hyper") diff --git a/internal/config/reload_hooks_test.go b/internal/config/reload_hooks_test.go new file mode 100644 index 0000000000000000000000000000000000000000..a21dfa89a0b7c61260a96f800acc34a18a2b086b --- /dev/null +++ b/internal/config/reload_hooks_test.go @@ -0,0 +1,102 @@ +package config_test + +import ( + "context" + "os" + "path/filepath" + "testing" + + "github.com/charmbracelet/crush/internal/config" + "github.com/charmbracelet/crush/internal/hooks" + "github.com/stretchr/testify/require" +) + +// TestReloadFromDisk_CompilesHookMatchers is a regression test for a bug +// where ReloadFromDisk dropped the compiled matcher regex on every hook, +// causing a matcher like "^bash$" to match every tool call after any +// SetConfigField-triggered reload. +// +// The assertion is phrased in terms of observable Runner behavior (not +// internal field presence) so it stays valid if the Runner later owns +// matcher compilation itself. +func TestReloadFromDisk_CompilesHookMatchers(t *testing.T) { + // No t.Parallel(): we Setenv HOME/XDG_CONFIG_HOME to isolate from the + // developer's real global config, which may define its own hooks. + isolated := t.TempDir() + t.Setenv("HOME", isolated) + t.Setenv("XDG_CONFIG_HOME", filepath.Join(isolated, ".config")) + t.Setenv("XDG_DATA_HOME", filepath.Join(isolated, ".local", "share")) + + workDir := t.TempDir() + dataDir := t.TempDir() + configPath := filepath.Join(workDir, "crush.json") + cfgJSON := `{ + "hooks": { + "PreToolUse": [ + {"matcher": "^bash$", "command": "exit 0"} + ] + } + }` + require.NoError(t, os.WriteFile(configPath, []byte(cfgJSON), 0o600)) + + store, err := config.Load(workDir, dataDir, false) + require.NoError(t, err) + + // Sanity: hook filtering works immediately after Load. + assertHookFilters(t, store) + + require.NoError(t, store.ReloadFromDisk(context.Background())) + + // The actual regression check: filtering must still work after a + // reload, not silently collapse to match-everything. + assertHookFilters(t, store) +} + +// assertHookFilters builds a Runner from the store's current hooks and +// verifies the "^bash$" matcher rejects a non-bash tool while accepting +// bash. +func assertHookFilters(t *testing.T, store *config.ConfigStore) { + t.Helper() + preHooks := store.Config().Hooks[hooks.EventPreToolUse] + require.Len(t, preHooks, 1) + + runner := hooks.NewRunner(preHooks, t.TempDir(), t.TempDir()) + + nonMatch, err := runner.Run(context.Background(), hooks.EventPreToolUse, "sess", "view", `{}`) + require.NoError(t, err) + require.Equal(t, 0, nonMatch.HookCount, "view must not match ^bash$ matcher") + + match, err := runner.Run(context.Background(), hooks.EventPreToolUse, "sess", "bash", `{}`) + require.NoError(t, err) + require.Equal(t, 1, match.HookCount, "bash must match ^bash$ matcher") +} + +// TestSetConfigField_AutoReload_PreservesHookMatcherFiltering verifies the +// dominant real-world trigger path: config writes call autoReload, +// autoReload calls ReloadFromDisk, and hook matching must remain correct. +func TestSetConfigField_AutoReload_PreservesHookMatcherFiltering(t *testing.T) { + isolated := t.TempDir() + t.Setenv("HOME", isolated) + t.Setenv("XDG_CONFIG_HOME", filepath.Join(isolated, ".config")) + t.Setenv("XDG_DATA_HOME", filepath.Join(isolated, ".local", "share")) + + workDir := t.TempDir() + dataDir := t.TempDir() + configPath := filepath.Join(workDir, "crush.json") + cfgJSON := `{ + "hooks": { + "PreToolUse": [ + {"matcher": "^bash$", "command": "exit 0"} + ] + } + }` + require.NoError(t, os.WriteFile(configPath, []byte(cfgJSON), 0o600)) + + store, err := config.Load(workDir, dataDir, false) + require.NoError(t, err) + assertHookFilters(t, store) + + require.NoError(t, store.SetConfigField(config.ScopeGlobal, "options.debug", true)) + + assertHookFilters(t, store) +} diff --git a/internal/config/store.go b/internal/config/store.go index 86fa937d1595b83f97fa43d3abd0a39cabd5f8cc..c44fceba0ac87133a787b1351de213802a4e17cc 100644 --- a/internal/config/store.go +++ b/internal/config/store.go @@ -3,6 +3,7 @@ package config import ( "cmp" "context" + "encoding/json" "fmt" "log/slog" "os" @@ -288,6 +289,9 @@ func (s *ConfigStore) SetProviderAPIKey(scope Scope, providerID string, apiKey a } // RefreshOAuthToken refreshes the OAuth token for the given provider. +// Before making an external refresh request, it checks the config file on +// disk to see if another Crush session has already refreshed the token. If +// a newer token is found, it is used instead of refreshing. func (s *ConfigStore) RefreshOAuthToken(ctx context.Context, scope Scope, providerID string) error { providerConfig, exists := s.config.Providers.Get(providerID) if !exists { @@ -298,13 +302,29 @@ func (s *ConfigStore) RefreshOAuthToken(ctx context.Context, scope Scope, provid return fmt.Errorf("provider %s does not have an OAuth token", providerID) } - var newToken *oauth.Token + // Check if another session refreshed the token recently by reading + // the current token from the config file on disk. + newToken, err := s.loadTokenFromDisk(scope, providerID) + if err != nil { + slog.Warn("Failed to read token from config file, proceeding with refresh", "provider", providerID, "error", err) + } else if newToken != nil && newToken.AccessToken != providerConfig.OAuthToken.AccessToken { + slog.Info("Using token refreshed by another session", "provider", providerID) + providerConfig.OAuthToken = newToken + providerConfig.APIKey = newToken.AccessToken + if providerID == string(catwalk.InferenceProviderCopilot) { + providerConfig.SetupGitHubCopilot() + } + s.config.Providers.Set(providerID, providerConfig) + return nil + } + + var refreshedToken *oauth.Token var refreshErr error switch providerID { case string(catwalk.InferenceProviderCopilot): - newToken, refreshErr = copilot.RefreshToken(ctx, providerConfig.OAuthToken.RefreshToken) + refreshedToken, refreshErr = copilot.RefreshToken(ctx, providerConfig.OAuthToken.RefreshToken) case hyperp.Name: - newToken, refreshErr = hyper.ExchangeToken(ctx, providerConfig.OAuthToken.RefreshToken) + refreshedToken, refreshErr = hyper.ExchangeToken(ctx, providerConfig.OAuthToken.RefreshToken) default: return fmt.Errorf("OAuth refresh not supported for provider %s", providerID) } @@ -313,8 +333,8 @@ func (s *ConfigStore) RefreshOAuthToken(ctx context.Context, scope Scope, provid } slog.Info("Successfully refreshed OAuth token", "provider", providerID) - providerConfig.OAuthToken = newToken - providerConfig.APIKey = newToken.AccessToken + providerConfig.OAuthToken = refreshedToken + providerConfig.APIKey = refreshedToken.AccessToken switch providerID { case string(catwalk.InferenceProviderCopilot): @@ -324,8 +344,8 @@ func (s *ConfigStore) RefreshOAuthToken(ctx context.Context, scope Scope, provid s.config.Providers.Set(providerID, providerConfig) if err := cmp.Or( - s.SetConfigField(scope, fmt.Sprintf("providers.%s.api_key", providerID), newToken.AccessToken), - s.SetConfigField(scope, fmt.Sprintf("providers.%s.oauth", providerID), newToken), + s.SetConfigField(scope, fmt.Sprintf("providers.%s.api_key", providerID), refreshedToken.AccessToken), + s.SetConfigField(scope, fmt.Sprintf("providers.%s.oauth", providerID), refreshedToken), ); err != nil { return fmt.Errorf("failed to persist refreshed token: %w", err) } @@ -333,6 +353,41 @@ func (s *ConfigStore) RefreshOAuthToken(ctx context.Context, scope Scope, provid return nil } +// loadTokenFromDisk reads the OAuth token for the given provider from the +// config file on disk. Returns nil if the token is not found or matches the +// current in-memory token. +func (s *ConfigStore) loadTokenFromDisk(scope Scope, providerID string) (*oauth.Token, error) { + path, err := s.configPath(scope) + if err != nil { + return nil, err + } + + data, err := os.ReadFile(path) + if err != nil { + if os.IsNotExist(err) { + return nil, nil + } + return nil, err + } + + oauthKey := fmt.Sprintf("providers.%s.oauth", providerID) + oauthResult := gjson.Get(string(data), oauthKey) + if !oauthResult.Exists() { + return nil, nil + } + + var token oauth.Token + if err := json.Unmarshal([]byte(oauthResult.Raw), &token); err != nil { + return nil, err + } + + if token.AccessToken == "" { + return nil, nil + } + + return &token, nil +} + // recordRecentModel records a model in the recent models list. func (s *ConfigStore) recordRecentModel(scope Scope, modelType SelectedModelType, model SelectedModel) error { if model.Provider == "" || model.Model == "" { @@ -590,6 +645,12 @@ func (s *ConfigStore) ReloadFromDisk(ctx context.Context) error { } } + // Validate hooks after all config merging is complete so matcher + // regexes are recompiled on the reloaded config (mirrors Load). + if err := cfg.ValidateHooks(); err != nil { + return fmt.Errorf("invalid hook configuration on reload: %w", err) + } + // Preserve runtime overrides overrides := s.overrides diff --git a/internal/config/store_test.go b/internal/config/store_test.go index 46d51440870c0361b6e0246881d4522fa044363c..3f9441ea02315a6ac75012d956acbcab217d1c91 100644 --- a/internal/config/store_test.go +++ b/internal/config/store_test.go @@ -8,6 +8,8 @@ import ( "testing" "time" + "github.com/charmbracelet/crush/internal/csync" + "github.com/charmbracelet/crush/internal/oauth" "github.com/stretchr/testify/require" ) @@ -510,3 +512,183 @@ func TestAutoReloadDisabledDuringReload(t *testing.T) { // Verify reload completed successfully require.False(t, store.autoReloadDisabled, "autoReloadDisabled should be false after ReloadFromDisk") } + +func TestLoadTokenFromDisk_ReturnsNewerToken(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "crush.json") + + // Create config file with a newer token on disk + configContent := `{ + "providers": { + "hyper": { + "oauth": { + "access_token": "newer-token-from-disk", + "refresh_token": "refresh-abc", + "expires_in": 3600, + "expires_at": 9999999999 + } + } + } + }` + require.NoError(t, os.WriteFile(configPath, []byte(configContent), 0o600)) + + store := &ConfigStore{ + config: &Config{}, + globalDataPath: configPath, + } + + token, err := store.loadTokenFromDisk(ScopeGlobal, "hyper") + require.NoError(t, err) + require.NotNil(t, token) + require.Equal(t, "newer-token-from-disk", token.AccessToken) + require.Equal(t, "refresh-abc", token.RefreshToken) + require.Equal(t, 3600, token.ExpiresIn) + require.Equal(t, int64(9999999999), token.ExpiresAt) +} + +func TestLoadTokenFromDisk_ReturnsNilWhenSameToken(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "crush.json") + + // Create config file with the same token + configContent := `{ + "providers": { + "hyper": { + "oauth": { + "access_token": "same-token", + "refresh_token": "refresh-abc", + "expires_in": 3600, + "expires_at": 9999999999 + } + } + } + }` + require.NoError(t, os.WriteFile(configPath, []byte(configContent), 0o600)) + + store := &ConfigStore{ + config: &Config{}, + globalDataPath: configPath, + } + + token, err := store.loadTokenFromDisk(ScopeGlobal, "hyper") + require.NoError(t, err) + require.NotNil(t, token) + require.Equal(t, "same-token", token.AccessToken) +} + +func TestLoadTokenFromDisk_ReturnsNilWhenFileMissing(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "nonexistent.json") + + store := &ConfigStore{ + config: &Config{}, + globalDataPath: configPath, + } + + token, err := store.loadTokenFromDisk(ScopeGlobal, "hyper") + require.NoError(t, err) + require.Nil(t, token) +} + +func TestLoadTokenFromDisk_ReturnsNilWhenProviderMissing(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "crush.json") + + // Create config file without the hyper provider + configContent := `{"providers": {"openai": {"api_key": "test-key"}}}` + require.NoError(t, os.WriteFile(configPath, []byte(configContent), 0o600)) + + store := &ConfigStore{ + config: &Config{}, + globalDataPath: configPath, + } + + token, err := store.loadTokenFromDisk(ScopeGlobal, "hyper") + require.NoError(t, err) + require.Nil(t, token) +} + +func TestLoadTokenFromDisk_ReturnsNilWhenOAuthMissing(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "crush.json") + + // Create config file with provider but no OAuth token + configContent := `{"providers": {"hyper": {"api_key": "test-key"}}}` + require.NoError(t, os.WriteFile(configPath, []byte(configContent), 0o600)) + + store := &ConfigStore{ + config: &Config{}, + globalDataPath: configPath, + } + + token, err := store.loadTokenFromDisk(ScopeGlobal, "hyper") + require.NoError(t, err) + require.Nil(t, token) +} + +func TestRefreshOAuthToken_UsesDiskTokenWhenDifferent(t *testing.T) { + t.Parallel() + + dir := t.TempDir() + configPath := filepath.Join(dir, "crush.json") + + // Create config file with a newer token on disk + configContent := `{ + "providers": { + "hyper": { + "api_key": "newer-access-token", + "oauth": { + "access_token": "newer-access-token", + "refresh_token": "refresh-abc", + "expires_in": 3600, + "expires_at": 9999999999 + } + } + } + }` + require.NoError(t, os.WriteFile(configPath, []byte(configContent), 0o600)) + + // Set up store with an older in-memory token + oldToken := &oauth.Token{ + AccessToken: "older-access-token", + RefreshToken: "refresh-abc", + ExpiresIn: 3600, + ExpiresAt: time.Now().Add(-time.Hour).Unix(), // Expired + } + + providers := csync.NewMap[string, ProviderConfig]() + providers.Set("hyper", ProviderConfig{ + ID: "hyper", + Name: "Hyper", + APIKey: oldToken.AccessToken, + OAuthToken: oldToken, + }) + + store := &ConfigStore{ + config: &Config{ + Providers: providers, + }, + globalDataPath: configPath, + } + + // Refresh should use the disk token without making an external call + err := store.RefreshOAuthToken(context.Background(), ScopeGlobal, "hyper") + require.NoError(t, err) + + // Verify the in-memory token was updated to the disk token + updatedConfig, ok := store.config.Providers.Get("hyper") + require.True(t, ok) + require.Equal(t, "newer-access-token", updatedConfig.APIKey) + require.Equal(t, "newer-access-token", updatedConfig.OAuthToken.AccessToken) + require.Equal(t, "refresh-abc", updatedConfig.OAuthToken.RefreshToken) +} diff --git a/internal/db/connect.go b/internal/db/connect.go index e31d220c7454a3ccc4241b336bedbc7bf30d5fba..ef800c716efc44b137c163a188f599366f1c66e3 100644 --- a/internal/db/connect.go +++ b/internal/db/connect.go @@ -50,6 +50,13 @@ func Connect(ctx context.Context, dataDir string) (*sql.DB, error) { return nil, err } + // Serialize all access through a single connection. SQLite serializes + // writes at the file level anyway, and allowing multiple pool + // connections to interleave writes/checkpoints (especially under + // concurrent sub-agents) has caused WAL/header desync resulting in + // SQLITE_NOTADB (26) on the next open. + db.SetMaxOpenConns(1) + if err = db.PingContext(ctx); err != nil { db.Close() return nil, fmt.Errorf("failed to connect to database: %w", err) diff --git a/internal/db/connect_modernc.go b/internal/db/connect_modernc.go index 39c7faa42516297d4df497821baa0be56835be15..2e9676d6420c6fbda69cb924595a6e8a55740a5c 100644 --- a/internal/db/connect_modernc.go +++ b/internal/db/connect_modernc.go @@ -17,6 +17,9 @@ func openDB(dbPath string) (*sql.DB, error) { for name, value := range pragmas { params.Add("_pragma", fmt.Sprintf("%s(%s)", name, value)) } + // Use BEGIN IMMEDIATE so writers acquire the reserved lock up front, + // preventing deferred-to-writer upgrade deadlocks. + params.Set("_txlock", "immediate") dsn := fmt.Sprintf("file:%s?%s", dbPath, params.Encode()) db, err := sql.Open("sqlite", dsn) diff --git a/internal/db/connect_ncruces.go b/internal/db/connect_ncruces.go index 12946e29439862653e0253adc9eb6e2869bea559..d90833350030f08be364f940d483e0afb0b63eb5 100644 --- a/internal/db/connect_ncruces.go +++ b/internal/db/connect_ncruces.go @@ -11,7 +11,11 @@ import ( ) func openDB(dbPath string) (*sql.DB, error) { - db, err := driver.Open(dbPath, func(c *sqlite3.Conn) error { + // Use BEGIN IMMEDIATE so writers acquire the reserved lock up front, + // preventing deferred-to-writer upgrade deadlocks. The "file:" prefix + // is required for the ncruces driver to parse query parameters. + dsn := fmt.Sprintf("file:%s?_txlock=immediate", dbPath) + db, err := driver.Open(dsn, func(c *sqlite3.Conn) error { // Set pragmas for better performance via _pragma query params. // Format: PRAGMA name = value; for name, value := range pragmas { diff --git a/internal/hooks/hooks_test.go b/internal/hooks/hooks_test.go index 79236aaf2479557a59a93267c0b3695001cf9971..77217e77354367a07d8a2a8e172152b2281e3643 100644 --- a/internal/hooks/hooks_test.go +++ b/internal/hooks/hooks_test.go @@ -383,6 +383,40 @@ func TestRunnerMatcherFiltering(t *testing.T) { require.NoError(t, err) require.Equal(t, DecisionNone, result.Decision) }) + + // Runner must compile matchers itself; it cannot rely on + // ValidateHooks having run first. This is the guarantee that prevents + // the reload-drops-matcher class of bug. + t.Run("runner compiles matcher without ValidateHooks", func(t *testing.T) { + t.Parallel() + raw := []config.HookConfig{ + {Command: `echo '{"decision":"deny","reason":"blocked"}'`, Matcher: "^bash$"}, + } + r := NewRunner(raw, t.TempDir(), t.TempDir()) + + deny, err := r.Run(context.Background(), EventPreToolUse, "sess", "bash", `{}`) + require.NoError(t, err) + require.Equal(t, DecisionDeny, deny.Decision) + + noop, err := r.Run(context.Background(), EventPreToolUse, "sess", "view", `{}`) + require.NoError(t, err) + require.Equal(t, DecisionNone, noop.Decision) + }) + + // A matcher that fails to compile at Runner construction must not + // degrade to match-everything; the hook is dropped instead. + t.Run("runner skips hooks with invalid matcher", func(t *testing.T) { + t.Parallel() + raw := []config.HookConfig{ + {Command: `echo '{"decision":"deny","reason":"should not fire"}'`, Matcher: "[invalid"}, + } + r := NewRunner(raw, t.TempDir(), t.TempDir()) + + result, err := r.Run(context.Background(), EventPreToolUse, "sess", "bash", `{}`) + require.NoError(t, err) + require.Equal(t, DecisionNone, result.Decision) + require.Empty(t, r.Hooks()) + }) } func TestValidateHooksInvalidRegex(t *testing.T) { diff --git a/internal/hooks/runner.go b/internal/hooks/runner.go index e3bf6ca29c36ef35524d6b4d645b7112118c79e5..be625b0f10f3d97b4c67220dce3d96720797eb52 100644 --- a/internal/hooks/runner.go +++ b/internal/hooks/runner.go @@ -4,6 +4,7 @@ import ( "bytes" "context" "log/slog" + "regexp" "strings" "sync" "time" @@ -24,25 +25,63 @@ const abandonGrace = time.Second // depending on the scheduling behavior of the real interpreter. var runShell = shell.Run +// compiledHook pairs a HookConfig with its compiled matcher regex. A nil +// matcher means "match every tool". +type compiledHook struct { + cfg config.HookConfig + matcher *regexp.Regexp +} + // Runner executes hook commands and aggregates their results. type Runner struct { - hooks []config.HookConfig + hooks []compiledHook cwd string projectDir string } -// NewRunner creates a Runner from the given hook configs. +// NewRunner creates a Runner from the given hook configs. Each hook's +// Matcher is compiled here so the Runner is self-sufficient; callers do +// not have to pre-compile matchers on the config, and reloads or merges +// that rebuild HookConfig values can't silently strip compiled state. +// +// Hooks whose matcher fails to compile are skipped with a warning rather +// than treated as match-everything. ValidateHooks is expected to have +// caught syntax errors earlier, so this is defense in depth. func NewRunner(hooks []config.HookConfig, cwd, projectDir string) *Runner { + compiled := make([]compiledHook, 0, len(hooks)) + for _, h := range hooks { + ch := compiledHook{cfg: h} + if h.Matcher != "" { + re, err := regexp.Compile(h.Matcher) + if err != nil { + slog.Warn("Hook matcher failed to compile; skipping hook", + "matcher", h.Matcher, + "command", h.Command, + "error", err, + ) + continue + } + ch.matcher = re + } + compiled = append(compiled, ch) + } return &Runner{ - hooks: hooks, + hooks: compiled, cwd: cwd, projectDir: projectDir, } } -// Hooks returns the hook configs the runner was created with. +// Hooks returns the hook configs the runner was created with, in config +// order. Hooks whose matcher failed to compile at construction are +// omitted. Intended for diagnostics; callers should not rely on ordering +// or identity beyond that. func (r *Runner) Hooks() []config.HookConfig { - return r.hooks + out := make([]config.HookConfig, len(r.hooks)) + for i, h := range r.hooks { + out[i] = h.cfg + } + return out } // Run executes all matching hooks for the given event and tool, returning @@ -105,9 +144,8 @@ func (r *Runner) Run(ctx context.Context, eventName, sessionID, toolName, toolIn func (r *Runner) matchingHooks(toolName string) []config.HookConfig { var matched []config.HookConfig for _, h := range r.hooks { - re := h.MatcherRegex() - if re == nil || re.MatchString(toolName) { - matched = append(matched, h) + if h.matcher == nil || h.matcher.MatchString(toolName) { + matched = append(matched, h.cfg) } } return matched diff --git a/internal/server/net_other.go b/internal/server/net_other.go index 1ef7066c13abc1ab94494641d29449823ea8dd15..ba84404fdbb257cc77354c842bebbab7ce1d156c 100644 --- a/internal/server/net_other.go +++ b/internal/server/net_other.go @@ -1,5 +1,4 @@ //go:build !windows -// +build !windows package server diff --git a/internal/server/proto.go b/internal/server/proto.go index af34131810a7af8c3672fe460198d25afe9ba064..af591173d64e4c56ae51964ca56fcfbbca9658a6 100644 --- a/internal/server/proto.go +++ b/internal/server/proto.go @@ -199,7 +199,7 @@ func (c *controllerV1) handleGetWorkspaceProviders(w http.ResponseWriter, r *htt func (c *controllerV1) handleGetWorkspaceEvents(w http.ResponseWriter, r *http.Request) { flusher := http.NewResponseController(w) id := r.PathValue("id") - events, err := c.backend.SubscribeEvents(id) + events, err := c.backend.SubscribeEvents(r.Context(), id) if err != nil { c.handleError(w, r, err) return @@ -218,8 +218,8 @@ func (c *controllerV1) handleGetWorkspaceEvents(w http.ResponseWriter, r *http.R if !ok { return } - c.server.logDebug(r, "Sending event", "event", fmt.Sprintf("%T %+v", ev, ev)) - wrapped := wrapEvent(ev) + c.server.logDebug(r, "Sending event", "event", fmt.Sprintf("%T %+v", ev.Payload, ev.Payload)) + wrapped := wrapEvent(ev.Payload) if wrapped == nil { continue } diff --git a/internal/ui/attachments/attachments.go b/internal/ui/attachments/attachments.go index 558c7576ee1edb3756be3dc7b4ccfcb89a5597b7..d56ea7ac43706ecf36b35fd9ec912d660370eaf1 100644 --- a/internal/ui/attachments/attachments.go +++ b/internal/ui/attachments/attachments.go @@ -78,6 +78,10 @@ func (m *Attachments) Render(width int) string { return m.renderer.Render(m.list, m.deleting, width) } +// Renderer returns the attachment renderer so callers can update its +// styles in place. +func (m *Attachments) Renderer() *Renderer { return m.renderer } + func NewRenderer(normalStyle, deletingStyle, imageStyle, textStyle lipgloss.Style) *Renderer { return &Renderer{ normalStyle: normalStyle, @@ -87,6 +91,14 @@ func NewRenderer(normalStyle, deletingStyle, imageStyle, textStyle lipgloss.Styl } } +// SetStyles updates the renderer styles in place. +func (r *Renderer) SetStyles(normalStyle, deletingStyle, imageStyle, textStyle lipgloss.Style) { + r.normalStyle = normalStyle + r.textStyle = textStyle + r.imageStyle = imageStyle + r.deletingStyle = deletingStyle +} + type Renderer struct { normalStyle, textStyle, imageStyle, deletingStyle lipgloss.Style } diff --git a/internal/ui/chat/messages.go b/internal/ui/chat/messages.go index 4906516cc1d259037f27675e92f60efa34fa817c..6c86cb74951e0bf8f20fe7af4fa420b4527936ca 100644 --- a/internal/ui/chat/messages.go +++ b/internal/ui/chat/messages.go @@ -128,6 +128,22 @@ func defaultHighlighter(sty *styles.Styles) *highlightableMessageItem { } } +// cacheClearable is implemented by message items that cache rendered +// output and can be asked to drop the cache. +type cacheClearable interface { + clearCache() +} + +// ClearItemCaches drops any cached rendered output on each item so the +// next render uses the current styles. +func ClearItemCaches(items []MessageItem) { + for _, item := range items { + if cc, ok := item.(cacheClearable); ok { + cc.clearCache() + } + } +} + // cachedMessageItem caches rendered message content to avoid re-rendering. // // This should be used by any message that can store a cached version of its render. e.x user,assistant... and so on diff --git a/internal/ui/chat/tools.go b/internal/ui/chat/tools.go index c0bc055e9d9a49341261576ef29df1f6cd33b546..4715cfafa6a25f01b93ee56caf41aa55aee3196e 100644 --- a/internal/ui/chat/tools.go +++ b/internal/ui/chat/tools.go @@ -797,22 +797,27 @@ func renderHookLine(sty *styles.Styles, hi hooks.HookInfo, rawName, detail strin // hookDetail returns the styled detail text for a single hook result. func hookDetail(sty *styles.Styles, hi hooks.HookInfo) string { + const ( + okMessage = "OK" + denialMessage = "Denied" + rewroteMessage = "Rewrote Output" + ) switch hi.Decision { case "deny": if hi.Reason != "" { - return sty.Tool.HookDenied.Render("Denied") + " " + sty.Tool.HookDeniedReason.Render(hi.Reason) + return sty.Tool.HookDenied.Render(denialMessage) + " " + sty.Tool.HookDeniedReason.Render(hi.Reason) } - return sty.Tool.HookDenied.Render("Denied") + return sty.Tool.HookDenied.Render(denialMessage) case "allow": - result := sty.Tool.HookOK.Render("OK") + result := sty.Tool.HookOK.Render(okMessage) if hi.InputRewrite { - result += " " + sty.Tool.HookRewrote.Render("Rewrote Input") + result += " " + sty.Tool.HookRewrote.Render(rewroteMessage) } return result default: - result := sty.Tool.HookOK.Render("OK") + result := sty.Tool.HookOK.Render(okMessage) if hi.InputRewrite { - result += " " + sty.Tool.HookRewrote.Render("Rewrote Input") + result += " " + sty.Tool.HookRewrote.Render(rewroteMessage) } return result } diff --git a/internal/ui/common/common.go b/internal/ui/common/common.go index 798ee2c1bf1d99aa15d4f343956bf16170c147f3..88d6f1f64312a48dde5474ac77e4bcef0a141b9a 100644 --- a/internal/ui/common/common.go +++ b/internal/ui/common/common.go @@ -31,15 +31,36 @@ func (c *Common) Config() *config.Config { return c.Workspace.Config() } -// DefaultCommon returns the default common UI configurations. +// DefaultCommon returns the default common UI configurations. When the +// workspace has a large model selected, the theme is chosen based on its +// provider; otherwise the default theme is used. func DefaultCommon(ws workspace.Workspace) *Common { - s := styles.CharmtonePantera() + s := styles.ThemeForProvider(largeModelProviderID(ws)) return &Common{ Workspace: ws, Styles: &s, } } +// largeModelProviderID returns the provider ID of the currently selected +// large model, or the empty string if none is set or the workspace is nil. +func largeModelProviderID(ws workspace.Workspace) string { + if ws == nil { + return "" + } + cfg := ws.Config() + if cfg == nil { + return "" + } + return cfg.Models[config.SelectedModelTypeLarge].Provider +} + +// IsHyper reports whether the currently selected large model is provided +// by Hyper. +func (c *Common) IsHyper() bool { + return largeModelProviderID(c.Workspace) == "hyper" +} + // CenterRect returns a new [Rectangle] centered within the given area with the // specified width and height. func CenterRect(area uv.Rectangle, width, height int) uv.Rectangle { diff --git a/internal/ui/common/elements.go b/internal/ui/common/elements.go index 652d4734345397cdbb3a7e3160f842ca6a5cca74..902c8a816247d97f0d9dd0ce7538b1bf130bc07d 100644 --- a/internal/ui/common/elements.go +++ b/internal/ui/common/elements.go @@ -4,9 +4,11 @@ import ( "cmp" "fmt" "image/color" + "strconv" "strings" "charm.land/lipgloss/v2" + "github.com/charmbracelet/crush/internal/agent/hyper" "github.com/charmbracelet/crush/internal/home" "github.com/charmbracelet/crush/internal/ui/styles" "github.com/charmbracelet/x/ansi" @@ -38,7 +40,7 @@ type ModelContextInfo struct { // ModelInfo renders model information including name, provider, reasoning // settings, and optional context usage/cost. -func ModelInfo(t *styles.Styles, modelName, providerName, reasoningInfo string, context *ModelContextInfo, width int) string { +func ModelInfo(t *styles.Styles, modelName, providerName, reasoningInfo string, context *ModelContextInfo, width int, hyperCredits *int) string { modelIcon := t.ModelInfo.Icon.Render(styles.ModelIcon) modelName = t.ModelInfo.Name.Render(modelName) @@ -76,6 +78,13 @@ func ModelInfo(t *styles.Styles, modelName, providerName, reasoningInfo string, parts = append(parts, lipgloss.NewStyle().PaddingLeft(2).Render(formattedInfo)) } + if providerName == hyper.DisplayName && hyperCredits != nil { + hcInfo := t.ModelInfo.HypercreditIcon.Render(styles.HypercreditIcon) + hcInfo += " " + hcInfo += t.ModelInfo.HypercreditText.Render(fmt.Sprintf("%s Hypercredits", FormatCredits(*hyperCredits))) + parts = append(parts, "", hcInfo) + } + return lipgloss.NewStyle().Width(width).Render( lipgloss.JoinVertical(lipgloss.Left, parts...), ) @@ -115,6 +124,28 @@ func formatTokensAndCost(t *styles.Styles, tokens, contextWindow int64, cost flo return fmt.Sprintf("%s %s", formattedTokens, formattedCost) } +// FormatCredits formats an integer with comma separators for thousands. +func FormatCredits(n int) string { + s := strconv.FormatInt(int64(n), 10) + if n < 1000 { + return s + } + // Calculate how many digits before the first comma. + firstGroup := len(s) % 3 + if firstGroup == 0 { + firstGroup = 3 + } + var b []byte + for i := 0; i < len(s); i++ { + if i > 0 && i == firstGroup { + b = append(b, ',') + firstGroup += 3 + } + b = append(b, s[i]) + } + return string(b) +} + // StatusOpts defines options for rendering a status line with icon, title, // description, and optional extra content. type StatusOpts struct { diff --git a/internal/ui/common/markdown.go b/internal/ui/common/markdown.go index 6123711d9a6353c5ee96c11765859479112aef70..8ba45bccd263350607763af4813f2b6b047a1d7f 100644 --- a/internal/ui/common/markdown.go +++ b/internal/ui/common/markdown.go @@ -2,6 +2,7 @@ package common import ( "image/color" + "sync" "charm.land/glamour/v2" "github.com/alecthomas/chroma/v2/formatters" @@ -18,24 +19,55 @@ func init() { formatters.Register(formatterName, xchroma.Formatter(zero, nil)) } +var ( + mdCacheMu sync.Mutex + mdCache = map[int]*glamour.TermRenderer{} + quietMDCache = map[int]*glamour.TermRenderer{} +) + // MarkdownRenderer returns a glamour [glamour.TermRenderer] configured with -// the given styles and width. +// the given styles and width. Renderers are memoized per width and shared +// across callers; call InvalidateMarkdownRendererCache when the active +// styles change. func MarkdownRenderer(sty *styles.Styles, width int) *glamour.TermRenderer { + mdCacheMu.Lock() + defer mdCacheMu.Unlock() + if r, ok := mdCache[width]; ok { + return r + } r, _ := glamour.NewTermRenderer( glamour.WithStyles(sty.Markdown), glamour.WithWordWrap(width), glamour.WithChromaFormatter(formatterName), ) + mdCache[width] = r return r } // QuietMarkdownRenderer returns a glamour [glamour.TermRenderer] with no colors -// (plain text with structure) and the given width. +// (plain text with structure) and the given width. Renderers are memoized per +// width and shared across callers. func QuietMarkdownRenderer(sty *styles.Styles, width int) *glamour.TermRenderer { + mdCacheMu.Lock() + defer mdCacheMu.Unlock() + if r, ok := quietMDCache[width]; ok { + return r + } r, _ := glamour.NewTermRenderer( glamour.WithStyles(sty.QuietMarkdown), glamour.WithWordWrap(width), glamour.WithChromaFormatter(formatterName), ) + quietMDCache[width] = r return r } + +// InvalidateMarkdownRendererCache drops every cached renderer. Call this +// whenever the active styles change so subsequent renderers pick up the new +// ansi.StyleConfig. +func InvalidateMarkdownRendererCache() { + mdCacheMu.Lock() + defer mdCacheMu.Unlock() + mdCache = map[int]*glamour.TermRenderer{} + quietMDCache = map[int]*glamour.TermRenderer{} +} diff --git a/internal/ui/completions/completions.go b/internal/ui/completions/completions.go index 0a63e789b040f3b17b671c1f6ddb5c8be0f12f1c..9393cce52410884a7e410588adccf1863f4aed3b 100644 --- a/internal/ui/completions/completions.go +++ b/internal/ui/completions/completions.go @@ -110,6 +110,15 @@ func New(normalStyle, focusedStyle, matchStyle lipgloss.Style) *Completions { } } +// SetStyles updates the styles used when rendering completion items. +// Existing items are not restyled; subsequent SetItems calls pick up the +// new styles. +func (c *Completions) SetStyles(normalStyle, focusedStyle, matchStyle lipgloss.Style) { + c.normalStyle = normalStyle + c.focusedStyle = focusedStyle + c.matchStyle = matchStyle +} + // IsOpen returns whether the completions popup is open. func (c *Completions) IsOpen() bool { return c.open diff --git a/internal/ui/dialog/oauth.go b/internal/ui/dialog/oauth.go index 25fa08dbe24c4cbd6f7afd3909b399ee7d0974d0..15444db26c8babaf91a76f45f83afa656222e2f0 100644 --- a/internal/ui/dialog/oauth.go +++ b/internal/ui/dialog/oauth.go @@ -222,7 +222,7 @@ func (m *OAuth) headerContent() string { textStyle = t.Dialog.PrimaryText dialogStyle = t.Dialog.View.Width(m.width) headerOffset = titleStyle.GetHorizontalFrameSize() + dialogStyle.GetHorizontalFrameSize() - dialogTitle = fmt.Sprintf("Authenticate with %s", m.oAuthProvider.name()) + dialogTitle = fmt.Sprintf("Let’s authenticate with %s", m.oAuthProvider.name()) ) if m.isOnboarding { return textStyle.Render(dialogTitle) @@ -232,13 +232,13 @@ func (m *OAuth) headerContent() string { func (m *OAuth) innerDialogContent() string { var ( - t = m.com.Styles - whiteStyle = t.Dialog.OAuth.Instructions - primaryStyle = t.Dialog.OAuth.Enter - greenStyle = t.Dialog.OAuth.Success - linkStyle = t.Dialog.OAuth.Link - errorStyle = t.Dialog.OAuth.ErrorText - mutedStyle = t.Dialog.OAuth.StatusText + t = m.com.Styles + instructionStyle = t.Dialog.OAuth.Instructions + enterKeyStyle = t.Dialog.OAuth.Enter + successStyle = t.Dialog.OAuth.Success + linkStyle = t.Dialog.OAuth.Link + errorStyle = t.Dialog.OAuth.ErrorText + statusTextStyle = t.Dialog.OAuth.StatusText ) switch m.State { @@ -248,8 +248,8 @@ func (m *OAuth) innerDialogContent() string { Width(m.width - 2). Align(lipgloss.Center). Render( - greenStyle.Render(m.spinner.View()) + - mutedStyle.Render("Initializing..."), + successStyle.Render(m.spinner.View()) + + statusTextStyle.Render("Initializing..."), ) case OAuthStateDisplay: @@ -257,9 +257,9 @@ func (m *OAuth) innerDialogContent() string { Margin(0, 1). Width(m.width - 2). Render( - whiteStyle.Render("Press ") + - primaryStyle.Render("enter") + - whiteStyle.Render(" to copy the code below and open the browser."), + instructionStyle.Render("Press ") + + enterKeyStyle.Render("enter") + + instructionStyle.Render(" to copy the code below and open the browser."), ) codeBox := lipgloss.NewStyle(). @@ -273,16 +273,16 @@ func (m *OAuth) innerDialogContent() string { ) link := linkStyle.Hyperlink(m.verificationURL, "id=oauth-verify").Render(m.verificationURL) - url := mutedStyle. + url := statusTextStyle. Margin(0, 1). Width(m.width - 2). - Render("Browser not opening? Refer to\n" + link) + Render("Browser not opening? Pay a visit to:\n" + link) waiting := lipgloss.NewStyle(). Margin(0, 1). Width(m.width - 2). Render( - greenStyle.Render(m.spinner.View()) + mutedStyle.Render("Verifying..."), + successStyle.Render(m.spinner.View()) + statusTextStyle.Render("Verifying..."), ) return lipgloss.JoinVertical( @@ -299,7 +299,7 @@ func (m *OAuth) innerDialogContent() string { ) case OAuthStateSuccess: - return greenStyle. + return successStyle. Margin(1). Width(m.width - 2). Render("Authentication successful!") diff --git a/internal/ui/logo/logo.go b/internal/ui/logo/logo.go index fa7aa6f8c1548eb29d7f99352e2f23f31612e90c..dbfca152f339b0024cd1e18931a3785d7d9c52e7 100644 --- a/internal/ui/logo/logo.go +++ b/internal/ui/logo/logo.go @@ -148,10 +148,18 @@ func Render(base lipgloss.Style, version string, compact bool, o Opts) string { // SmallRender renders a smaller version of the Crush logo, suitable for // smaller windows or sidebar usage. -func SmallRender(t *styles.Styles, width int) string { - title := t.Logo.SmallCharm.Render("Charm™") - title = fmt.Sprintf("%s %s", title, styles.ApplyBoldForegroundGrad(t.Logo.GradCanvas, "Crush", t.Logo.SmallGradFromColor, t.Logo.SmallGradToColor)) - remainingWidth := width - lipgloss.Width(title) - 1 // 1 for the space after "Crush" +func SmallRender(t *styles.Styles, width int, o Opts) string { + name := "Crush" + if o.Hyper { + name = "HYPERCRUSH" + } + charm := "Charm™" + if !o.Hyper { + charm = " " + charm + } + title := t.Logo.SmallCharm.Render(charm) + title = fmt.Sprintf("%s %s", title, styles.ApplyBoldForegroundGrad(t.Logo.GradCanvas, name, t.Logo.SmallGradFromColor, t.Logo.SmallGradToColor)) + remainingWidth := width - lipgloss.Width(title) - 1 // 1 for the space after the name if remainingWidth > 0 { lines := strings.Repeat("╱", remainingWidth) title = fmt.Sprintf("%s %s", title, t.Logo.SmallDiagonals.Render(lines)) diff --git a/internal/ui/model/chat.go b/internal/ui/model/chat.go index ccd2325507545b35c9ee2e664cd869da9d2a8a4f..5cc620102febbe4af724ee1770f93f0f8de13212 100644 --- a/internal/ui/model/chat.go +++ b/internal/ui/model/chat.go @@ -107,6 +107,18 @@ func (m *Chat) Len() int { return m.list.Len() } +// InvalidateRenderCaches drops cached rendered output on every message +// item so the next draw re-renders with the current styles. +func (m *Chat) InvalidateRenderCaches() { + items := make([]chat.MessageItem, 0, m.list.Len()) + for i := range m.list.Len() { + if item, ok := m.list.ItemAt(i).(chat.MessageItem); ok { + items = append(items, item) + } + } + chat.ClearItemCaches(items) +} + // SetMessages sets the chat messages to the provided list of message items. func (m *Chat) SetMessages(msgs ...chat.MessageItem) { m.idInxMap = make(map[string]int) diff --git a/internal/ui/model/header.go b/internal/ui/model/header.go index d3b728738cf622fe966d43fde566c29495d4d69a..1fee1eb05942be52781173e8d63fab8de34a2f9b 100644 --- a/internal/ui/model/header.go +++ b/internal/ui/model/header.go @@ -37,12 +37,30 @@ func newHeader(com *common.Common) *header { h := &header{ com: com, } - t := com.Styles - h.compactLogo = t.Header.Charm.Render("Charm™") + " " + - styles.ApplyBoldForegroundGrad(t.Header.LogoGradCanvas, "CRUSH", t.Header.LogoGradFromColor, t.Header.LogoGradToColor) + " " + h.refresh() return h } +// refresh rebuilds cached logo strings using the current styles. Call +// after the theme changes. +func (h *header) refresh() { + t := h.com.Styles + isHyper := h.com.IsHyper() + charm := "Charm™" + if !isHyper { + charm = " " + charm + } + name := "CRUSH" + if isHyper { + name = "HYPERCRUSH" + } + h.compactLogo = t.Header.Charm.Render(charm) + " " + + styles.ApplyBoldForegroundGrad(t.Header.LogoGradCanvas, name, t.Header.LogoGradFromColor, t.Header.LogoGradToColor) + " " + // Force drawHeader to re-render the wide logo on the next frame. + h.width = 0 + h.logo = "" +} + // drawHeader draws the header for the given session. func (h *header) drawHeader( scr uv.Screen, @@ -51,10 +69,11 @@ func (h *header) drawHeader( compact bool, detailsOpen bool, width int, + hyperCredits *int, ) { t := h.com.Styles if width != h.width || compact != h.compact { - h.logo = renderLogo(h.com.Styles, compact, width) + h.logo = renderLogo(h.com.Styles, compact, h.com.IsHyper(), width) } h.width = width @@ -83,6 +102,7 @@ func (h *header) drawHeader( lspErrorCount, detailsOpen, availDetailWidth, + hyperCredits, ) remainingWidth := width - @@ -113,6 +133,7 @@ func renderHeaderDetails( lspErrorCount int, detailsOpen bool, availWidth int, + hyperCredits *int, ) string { t := com.Styles @@ -130,6 +151,11 @@ func renderHeaderDetails( parts = append(parts, formattedPercentage) } + if com.IsHyper() && hyperCredits != nil { + hc := t.Header.Hypercredit.Render(styles.HypercreditIcon) + " " + t.Header.Percentage.Render(common.FormatCredits(*hyperCredits)) + parts = append(parts, hc) + } + const keystroke = "ctrl+d" if detailsOpen { parts = append(parts, t.Header.Keystroke.Render(keystroke)+t.Header.KeystrokeTip.Render(" close")) diff --git a/internal/ui/model/sidebar.go b/internal/ui/model/sidebar.go index 5e04b45badcdad7016f236ed1362e4e9c99441d5..2f548628f24168bbe5f7d4d8390fc24e472f9559 100644 --- a/internal/ui/model/sidebar.go +++ b/internal/ui/model/sidebar.go @@ -53,7 +53,7 @@ func (m *UI) modelInfo(width int) string { if model != nil { modelName = model.CatwalkCfg.Name } - return common.ModelInfo(m.com.Styles, modelName, providerName, reasoningInfo, modelContext, width) + return common.ModelInfo(m.com.Styles, modelName, providerName, reasoningInfo, modelContext, width, m.hyperCredits) } // getDynamicHeightLimits will give us the num of items to show in each section based on the height @@ -142,7 +142,9 @@ func (m *UI) drawSidebar(scr uv.Screen, area uv.Rectangle) { cwd := common.PrettyPath(t, m.com.Workspace.WorkingDir(), width) sidebarLogo := m.sidebarLogo if height < logoHeightBreakpoint { - sidebarLogo = logo.SmallRender(m.com.Styles, width) + sidebarLogo = logo.SmallRender(m.com.Styles, width, logo.Opts{ + Hyper: m.com.IsHyper(), + }) } blocks := []string{ sidebarLogo, diff --git a/internal/ui/model/status.go b/internal/ui/model/status.go index ad6f0f81f776389b3aecfca560967924a47ace09..cf53294cb977866ee7020f7960f7c8464fa9089d 100644 --- a/internal/ui/model/status.go +++ b/internal/ui/model/status.go @@ -101,11 +101,13 @@ func (s *Status) Draw(scr uv.Screen, area uv.Rectangle) { ind := indStyle.String() indWidth := lipgloss.Width(ind) + msgPad := msgStyle.GetPaddingLeft() + msgStyle.GetPaddingRight() + avail := max(0, area.Dx()-indWidth-msgPad) msg := strings.Join(strings.Split(s.msg.Msg, "\n"), " ") - msgWidth := lipgloss.Width(msg) - msg = ansi.Truncate(msg, area.Dx()-indWidth-msgWidth, "…") - padWidth := max(0, area.Dx()-indWidth-msgWidth) - msg += strings.Repeat(" ", padWidth) + msg = ansi.Truncate(msg, avail, "…") + if w := lipgloss.Width(msg); w < avail { + msg += strings.Repeat(" ", avail-w) + } info := msgStyle.Render(msg) // Draw the info message over the help view diff --git a/internal/ui/model/ui.go b/internal/ui/model/ui.go index 2bbd59906440ab08036622e646655638255a17c9..bdf1f4baab76405600a53379c7455285e5d3984f 100644 --- a/internal/ui/model/ui.go +++ b/internal/ui/model/ui.go @@ -25,6 +25,7 @@ import ( tea "charm.land/bubbletea/v2" "charm.land/catwalk/pkg/catwalk" "charm.land/lipgloss/v2" + "github.com/charmbracelet/crush/internal/agent/hyper" "github.com/charmbracelet/crush/internal/agent/notify" agenttools "github.com/charmbracelet/crush/internal/agent/tools" "github.com/charmbracelet/crush/internal/agent/tools/mcp" @@ -137,6 +138,13 @@ type ( // closeDialogMsg is sent to close the current dialog. closeDialogMsg struct{} + // hyperRefreshDoneMsg is sent after a silent Hyper OAuth refresh + // finishes. It carries the original model-selection action so the + // selection can be resumed. + hyperRefreshDoneMsg struct { + action dialog.ActionSelectModel + } + // copyChatHighlightMsg is sent to copy the current chat highlight to clipboard. copyChatHighlightMsg struct{} @@ -144,6 +152,11 @@ type ( sessionFilesUpdatesMsg struct { sessionFiles []SessionFile } + // creditsUpdatedMsg is sent when the remaining Hyper credits have been + // fetched from the API. + creditsUpdatedMsg struct { + credits int + } ) // UI represents the main user interface model. @@ -257,6 +270,9 @@ type UI struct { // mouse highlighting related state lastClickTime time.Time + // hyperCredits is the remaining Hyper credits, updated after each prompt. + hyperCredits *int + // Prompt history for up/down navigation through previous messages. promptHistory struct { messages []string @@ -331,7 +347,7 @@ func New(com *common.Common, initialSessionID string, continueLast bool) *UI { status := NewStatus(com, ui) - ui.setEditorPrompt(false) + ui.setEditorPrompt(com.Workspace.PermissionSkipRequests()) ui.randomizePlaceholders() ui.textarea.Placeholder = ui.readyPlaceholder ui.status = status @@ -379,6 +395,9 @@ func (m *UI) Init() tea.Cmd { if cmd := m.loadInitialSession(); cmd != nil { cmds = append(cmds, cmd) } + if m.com.IsHyper() { + cmds = append(cmds, m.fetchHyperCredits()) + } return tea.Batch(cmds...) } @@ -843,6 +862,12 @@ func (m *UI) Update(msg tea.Msg) (tea.Model, tea.Cmd) { m.textarea.SetValue(msg.Text) m.textarea.MoveToEnd() cmds = append(cmds, m.updateTextareaWithPrevHeight(msg, prevHeight)) + case hyperRefreshDoneMsg: + if cmd := m.handleSelectModel(msg.action); cmd != nil { + cmds = append(cmds, cmd) + } + case creditsUpdatedMsg: + m.hyperCredits = &msg.credits case util.InfoMsg: if msg.Type == util.InfoTypeError { slog.Error("Error reported", "error", msg.Msg) @@ -853,6 +878,18 @@ func (m *UI) Update(msg tea.Msg) (tea.Model, tea.Cmd) { ttl = DefaultStatusTTL } cmds = append(cmds, clearInfoMsgCmd(ttl)) + case app.UpdateAvailableMsg: + text := fmt.Sprintf("Crush update available: v%s → v%s.", msg.CurrentVersion, msg.LatestVersion) + if msg.IsDevelopment { + text = fmt.Sprintf("This is a development version of Crush. The latest version is v%s.", msg.LatestVersion) + } + ttl := 10 * time.Second + m.status.SetInfoMsg(util.InfoMsg{ + Type: util.InfoTypeUpdate, + Msg: text, + TTL: ttl, + }) + cmds = append(cmds, clearInfoMsgCmd(ttl)) case util.ClearStatusMsg: m.status.ClearInfoMsg() case completions.CompletionItemsLoadedMsg: @@ -1429,66 +1466,8 @@ func (m *UI) handleDialogMsg(msg tea.Msg) tea.Cmd { m.dialog.CloseDialog(dialog.CommandsID) case dialog.ActionSelectModel: - if m.isAgentBusy() { - cmds = append(cmds, util.ReportWarn("Agent is busy, please wait...")) - break - } - - cfg := m.com.Config() - if cfg == nil { - cmds = append(cmds, util.ReportError(errors.New("configuration not found"))) - break - } - - var ( - providerID = msg.Model.Provider - isCopilot = providerID == string(catwalk.InferenceProviderCopilot) - isConfigured = func() bool { _, ok := cfg.Providers.Get(providerID); return ok } - ) - - // Attempt to import GitHub Copilot tokens from VSCode if available. - if isCopilot && !isConfigured() && !msg.ReAuthenticate { - m.com.Workspace.ImportCopilot() - } - - if !isConfigured() || msg.ReAuthenticate { - m.dialog.CloseDialog(dialog.ModelsID) - if cmd := m.openAuthenticationDialog(msg.Provider, msg.Model, msg.ModelType); cmd != nil { - cmds = append(cmds, cmd) - } - break - } - - if err := m.com.Workspace.UpdatePreferredModel(config.ScopeGlobal, msg.ModelType, msg.Model); err != nil { - cmds = append(cmds, util.ReportError(err)) - } else if _, ok := cfg.Models[config.SelectedModelTypeSmall]; !ok { - // Ensure small model is set is unset. - smallModel := m.com.Workspace.GetDefaultSmallModel(providerID) - if err := m.com.Workspace.UpdatePreferredModel(config.ScopeGlobal, config.SelectedModelTypeSmall, smallModel); err != nil { - cmds = append(cmds, util.ReportError(err)) - } - } - - cmds = append(cmds, func() tea.Msg { - if err := m.com.Workspace.UpdateAgentModel(context.TODO()); err != nil { - return util.ReportError(err) - } - - modelMsg := fmt.Sprintf("%s model changed to %s", msg.ModelType, msg.Model.Model) - - return util.NewInfoMsg(modelMsg) - }) - - m.dialog.CloseDialog(dialog.APIKeyInputID) - m.dialog.CloseDialog(dialog.OAuthID) - m.dialog.CloseDialog(dialog.ModelsID) - - if isOnboarding { - m.setState(uiLanding, uiFocusEditor) - m.com.Config().SetupAgents() - if err := m.com.Workspace.InitCoderAgent(context.TODO()); err != nil { - cmds = append(cmds, util.ReportError(err)) - } + if cmd := m.handleSelectModel(msg); cmd != nil { + cmds = append(cmds, cmd) } case dialog.ActionSelectReasoningEffort: if m.isAgentBusy() { @@ -1594,6 +1573,137 @@ func substituteArgs(content string, args map[string]string) string { return content } +// refreshHyperAndRetrySelect returns a command that silently refreshes +// the Hyper OAuth token and then re-runs the model selection. If the +// refresh fails, the selection resumes with ReAuthenticate set so the +// OAuth dialog opens. +func (m *UI) refreshHyperAndRetrySelect(msg dialog.ActionSelectModel) tea.Cmd { + return func() tea.Msg { + ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second) + defer cancel() + if err := m.com.Workspace.RefreshOAuthToken(ctx, config.ScopeGlobal, "hyper"); err != nil { + slog.Warn("Hyper OAuth refresh failed, requesting re-auth", "error", err) + msg.ReAuthenticate = true + } + return hyperRefreshDoneMsg{action: msg} + } +} + +// fetchHyperCredits returns a command that asynchronously fetches the +// remaining Hyper credits from the API. +func (m *UI) fetchHyperCredits() tea.Cmd { + return func() tea.Msg { + cfg := m.com.Config() + if cfg == nil { + return nil + } + providerCfg, ok := cfg.Providers.Get(hyper.Name) + if !ok { + return nil + } + apiKey, err := m.com.Workspace.Resolver().ResolveValue(providerCfg.APIKey) + if err != nil || apiKey == "" { + return nil + } + ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second) + defer cancel() + credits, err := hyper.FetchCredits(ctx, apiKey) + if err != nil { + slog.Error("Failed to fetch Hyper credits", "error", err) + return nil + } + return creditsUpdatedMsg{credits: credits} + } +} + +// handleSelectModel performs the model selection after any provider +// pre-checks (such as a silent Hyper OAuth refresh) have completed. +func (m *UI) handleSelectModel(msg dialog.ActionSelectModel) tea.Cmd { + var cmds []tea.Cmd + + if m.isAgentBusy() { + return util.ReportWarn("Agent is busy, please wait...") + } + + cfg := m.com.Config() + if cfg == nil { + return util.ReportError(errors.New("configuration not found")) + } + + var ( + providerID = msg.Model.Provider + isCopilot = providerID == string(catwalk.InferenceProviderCopilot) + isConfigured = func() bool { _, ok := cfg.Providers.Get(providerID); return ok } + isOnboarding = m.state == uiOnboarding + ) + + // For Hyper, if the stored OAuth token is expired, try a silent + // refresh before deciding whether the provider is configured. Keeps + // users from hitting a 401 on their first message after the + // short-lived access token ages out. + if !msg.ReAuthenticate && providerID == "hyper" { + if pc, ok := cfg.Providers.Get(providerID); ok && pc.OAuthToken != nil && pc.OAuthToken.IsExpired() { + return m.refreshHyperAndRetrySelect(msg) + } + } + + // Attempt to import GitHub Copilot tokens from VSCode if available. + if isCopilot && !isConfigured() && !msg.ReAuthenticate { + m.com.Workspace.ImportCopilot() + } + + if !isConfigured() || msg.ReAuthenticate { + m.dialog.CloseDialog(dialog.ModelsID) + if cmd := m.openAuthenticationDialog(msg.Provider, msg.Model, msg.ModelType); cmd != nil { + cmds = append(cmds, cmd) + } + return tea.Batch(cmds...) + } + + if err := m.com.Workspace.UpdatePreferredModel(config.ScopeGlobal, msg.ModelType, msg.Model); err != nil { + cmds = append(cmds, util.ReportError(err)) + } else { + if msg.ModelType == config.SelectedModelTypeLarge { + // Swap the theme live based on the newly selected large + // model's provider. + m.applyTheme(styles.ThemeForProvider(providerID)) + } + if _, ok := cfg.Models[config.SelectedModelTypeSmall]; !ok { + // Ensure small model is set is unset. + smallModel := m.com.Workspace.GetDefaultSmallModel(providerID) + if err := m.com.Workspace.UpdatePreferredModel(config.ScopeGlobal, config.SelectedModelTypeSmall, smallModel); err != nil { + cmds = append(cmds, util.ReportError(err)) + } + } + } + + cmds = append(cmds, func() tea.Msg { + if err := m.com.Workspace.UpdateAgentModel(context.TODO()); err != nil { + return util.ReportError(err) + } + + modelMsg := fmt.Sprintf("%s model changed to %s", msg.ModelType, msg.Model.Model) + + return util.NewInfoMsg(modelMsg) + }) + + m.dialog.CloseDialog(dialog.APIKeyInputID) + m.dialog.CloseDialog(dialog.OAuthID) + m.dialog.CloseDialog(dialog.ModelsID) + + if isOnboarding { + m.setState(uiLanding, uiFocusEditor) + m.com.Config().SetupAgents() + if err := m.com.Workspace.InitCoderAgent(context.TODO()); err != nil { + cmds = append(cmds, util.ReportError(err)) + } + } else if m.com.IsHyper() { + cmds = append(cmds, m.fetchHyperCredits()) + } + + return tea.Batch(cmds...) +} + func (m *UI) openAuthenticationDialog(provider catwalk.Provider, model config.SelectedModel, modelType config.SelectedModelType) tea.Cmd { var ( dlg dialog.Dialog @@ -1999,6 +2109,7 @@ func (m *UI) drawHeader(scr uv.Screen, area uv.Rectangle) { m.isCompact, m.detailsOpen, area.Dx(), + m.hyperCredits, ) } @@ -2977,7 +3088,37 @@ func (m *UI) renderEditorView(width int) string { // cacheSidebarLogo renders and caches the sidebar logo at the specified width. func (m *UI) cacheSidebarLogo(width int) { - m.sidebarLogo = renderLogo(m.com.Styles, true, width) + m.sidebarLogo = renderLogo(m.com.Styles, true, m.com.IsHyper(), width) +} + +// applyTheme replaces the active styles with the given theme, drops the +// shared markdown renderer cache, and refreshes every component that +// caches style data. +func (m *UI) applyTheme(s styles.Styles) { + *m.com.Styles = s + common.InvalidateMarkdownRendererCache() + m.refreshStyles() +} + +// refreshStyles pushes the current *m.com.Styles into every subcomponent +// that copies or pre-renders style-dependent values at construction time. +func (m *UI) refreshStyles() { + t := m.com.Styles + m.header.refresh() + if m.layout.sidebar.Dx() > 0 { + m.cacheSidebarLogo(m.layout.sidebar.Dx()) + } + m.textarea.SetStyles(t.Editor.Textarea) + m.completions.SetStyles(t.Completions.Normal, t.Completions.Focused, t.Completions.Match) + m.attachments.Renderer().SetStyles( + t.Attachments.Normal, + t.Attachments.Deleting, + t.Attachments.Image, + t.Attachments.Text, + ) + m.todoSpinner.Style = t.Pills.TodoSpinner + m.status.help.Styles = t.Help + m.chat.InvalidateRenderCaches() } // sendMessage sends a message with the given content and attachments. @@ -3257,10 +3398,15 @@ func (m *UI) handlePermissionNotification(notification permission.PermissionNoti func (m *UI) handleAgentNotification(n notify.Notification) tea.Cmd { switch n.Type { case notify.TypeAgentFinished: - return m.sendNotification(notification.Notification{ + var cmds []tea.Cmd + cmds = append(cmds, m.sendNotification(notification.Notification{ Title: "Crush is waiting...", Message: fmt.Sprintf("Agent's turn completed in \"%s\"", n.SessionTitle), - }) + })) + if m.com.IsHyper() { + cmds = append(cmds, m.fetchHyperCredits()) + } + return tea.Batch(cmds...) case notify.TypeReAuthenticate: return m.handleReAuthenticate(n.ProviderID) default: @@ -3656,7 +3802,7 @@ func (m *UI) disableDockerMCP() tea.Msg { } // renderLogo renders the Crush logo with the given styles and dimensions. -func renderLogo(t *styles.Styles, compact bool, width int) string { +func renderLogo(t *styles.Styles, compact, hyper bool, width int) string { return logo.Render(t.Logo.GradCanvas, version.Version, compact, logo.Opts{ FieldColor: t.Logo.FieldColor, TitleColorA: t.Logo.TitleColorA, @@ -3664,5 +3810,6 @@ func renderLogo(t *styles.Styles, compact bool, width int) string { CharmColor: t.Logo.CharmColor, VersionColor: t.Logo.VersionColor, Width: width, + Hyper: hyper, }) } diff --git a/internal/ui/styles/quickstyle.go b/internal/ui/styles/quickstyle.go index 0adb16ed2f89e29f9da6556dbea9ee0577032f68..7e01299788620df6c7bc9e25f913e8968c8fd110 100644 --- a/internal/ui/styles/quickstyle.go +++ b/internal/ui/styles/quickstyle.go @@ -14,119 +14,83 @@ import ( "github.com/charmbracelet/x/exp/charmtone" ) -// quickStyleOpts is the palette of colors used by quickStyle to build a -// complete Styles value. Each field maps to a semantic role in the UI. +// quickStyleOpts is the palette of colors used by quickStyle to simplify the +// process of building a theme. type quickStyleOpts struct { // Brand. primary color.Color secondary color.Color - tertiary color.Color + accent color.Color + keyword color.Color - // Foregrounds. - fgBase color.Color - fgMuted color.Color - fgHalfMuted color.Color - fgSubtle color.Color + // Default foreground and background colors. + fgBase color.Color + bgBase color.Color + + // Low-contrast dividers, separators, and rule lines. + separator color.Color + + fgSubtle color.Color + fgMoreSubtle color.Color + fgMostSubtle color.Color // Contrast pairings: foregrounds designed to sit on top of a // matching background role. onPrimary color.Color // foreground on primary backgrounds. - onAccent color.Color // foreground on saturated status/accent backgrounds. - - // Backgrounds. - bgBase color.Color - bgBaseLighter color.Color - bgSubtle color.Color - bgOverlay color.Color - - // Borders. - border color.Color - borderFocus color.Color - - // Status. - danger color.Color - error color.Color - warning color.Color - warningStrong color.Color - busy color.Color - info color.Color - infoSubtle color.Color - infoMuted color.Color - success color.Color - successSubtle color.Color - successMuted color.Color + + bgMostVisible color.Color + bgLessVisible color.Color + bgLeastVisible color.Color + + // Statuses. + destructive color.Color + error color.Color + warning color.Color + warningSubtle color.Color + busy color.Color + info color.Color + infoMoreSubtle color.Color + infoMostSubtle color.Color + success color.Color + successMoreSubtle color.Color + successMostSubtle color.Color } -// quickStyle builds a complete Styles value from a palette of semantic -// colors. Themes should populate quickStyleOpts and call this rather than -// re-implementing every style rule. +// quickStyle builds the default Styles (that is, the default theme, Charmtone +// Pantera) from a palette of semi-semanticly-named colors. // -// The idea here is that you can do most of the work with quickStyle, then -// add overrides as needed. +// The idea here is that you can do most of the work on a theme with quickStyle, +// then add overrides as needed. func quickStyle(o quickStyleOpts) Styles { var ( - primary = o.primary - secondary = o.secondary - tertiary = o.tertiary - - fgBase = o.fgBase - fgMuted = o.fgMuted - fgHalfMuted = o.fgHalfMuted - fgSubtle = o.fgSubtle - - onPrimary = o.onPrimary - onAccent = o.onAccent - - bgBase = o.bgBase - bgBaseLighter = o.bgBaseLighter - bgSubtle = o.bgSubtle - bgOverlay = o.bgOverlay - - border = o.border - borderFocus = o.borderFocus - - danger = o.danger - error = o.error - warning = o.warning - warningStrong = o.warningStrong - busy = o.busy - info = o.info - infoSubtle = o.infoSubtle - infoMuted = o.infoMuted - success = o.success - successSubtle = o.successSubtle - successMuted = o.successMuted - ) - - var ( - base = lipgloss.NewStyle().Foreground(fgBase) - muted = lipgloss.NewStyle().Foreground(fgMuted) - subtle = lipgloss.NewStyle().Foreground(fgSubtle) + base = lipgloss.NewStyle().Foreground(o.fgBase) + muted = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + subtle = lipgloss.NewStyle().Foreground(o.fgMostSubtle) s Styles ) - s.Background = bgBase + s.Background = o.bgBase // Populate color fields - s.WorkingGradFromColor = primary - s.WorkingGradToColor = secondary - s.WorkingLabelColor = fgBase + s.WorkingGradFromColor = o.primary + s.WorkingGradToColor = o.secondary + s.WorkingLabelColor = o.fgBase s.TextInput = textinput.Styles{ Focused: textinput.StyleState{ Text: base, - Placeholder: base.Foreground(fgSubtle), - Prompt: base.Foreground(tertiary), - Suggestion: base.Foreground(fgSubtle), + Placeholder: base.Foreground(o.fgMostSubtle), + Prompt: base.Foreground(o.accent), + Suggestion: base.Foreground(o.fgMostSubtle), }, Blurred: textinput.StyleState{ - Text: base.Foreground(fgMuted), - Placeholder: base.Foreground(fgSubtle), - Prompt: base.Foreground(fgMuted), - Suggestion: base.Foreground(fgSubtle), + Text: base.Foreground(o.fgMoreSubtle), + Placeholder: base.Foreground(o.fgMostSubtle), + Prompt: base.Foreground(o.fgMoreSubtle), + Suggestion: base.Foreground(o.fgMostSubtle), }, Cursor: textinput.CursorStyle{ - Color: secondary, + Color: o.secondary, Shape: tea.CursorBlock, Blink: true, }, @@ -136,23 +100,23 @@ func quickStyle(o quickStyleOpts) Styles { Focused: textarea.StyleState{ Base: base, Text: base, - LineNumber: base.Foreground(fgSubtle), + LineNumber: base.Foreground(o.fgMostSubtle), CursorLine: base, - CursorLineNumber: base.Foreground(fgSubtle), - Placeholder: base.Foreground(fgSubtle), - Prompt: base.Foreground(tertiary), + CursorLineNumber: base.Foreground(o.fgMostSubtle), + Placeholder: base.Foreground(o.fgMostSubtle), + Prompt: base.Foreground(o.accent), }, Blurred: textarea.StyleState{ Base: base, - Text: base.Foreground(fgMuted), - LineNumber: base.Foreground(fgMuted), + Text: base.Foreground(o.fgMoreSubtle), + LineNumber: base.Foreground(o.fgMoreSubtle), CursorLine: base, - CursorLineNumber: base.Foreground(fgMuted), - Placeholder: base.Foreground(fgSubtle), - Prompt: base.Foreground(fgMuted), + CursorLineNumber: base.Foreground(o.fgMoreSubtle), + Placeholder: base.Foreground(o.fgMostSubtle), + Prompt: base.Foreground(o.fgMoreSubtle), }, Cursor: textarea.CursorStyle{ - Color: secondary, + Color: o.secondary, Shape: tea.CursorBlock, Blink: true, }, @@ -163,7 +127,7 @@ func quickStyle(o quickStyleOpts) Styles { StylePrimitive: ansi.StylePrimitive{ // BlockPrefix: "\n", // BlockSuffix: "\n", - Color: hex(fgHalfMuted), + Color: hex(o.fgSubtle), }, // Margin: new(uint(defaultMargin)), }, @@ -178,7 +142,7 @@ func quickStyle(o quickStyleOpts) Styles { Heading: ansi.StyleBlock{ StylePrimitive: ansi.StylePrimitive{ BlockSuffix: "\n", - Color: hex(info), + Color: hex(o.info), Bold: new(true), }, }, @@ -186,8 +150,8 @@ func quickStyle(o quickStyleOpts) Styles { StylePrimitive: ansi.StylePrimitive{ Prefix: " ", Suffix: " ", - Color: hex(warning), - BackgroundColor: hex(primary), + Color: hex(o.warningSubtle), + BackgroundColor: hex(o.primary), Bold: new(true), }, }, @@ -214,7 +178,7 @@ func quickStyle(o quickStyleOpts) Styles { H6: ansi.StyleBlock{ StylePrimitive: ansi.StylePrimitive{ Prefix: "###### ", - Color: hex(successMuted), + Color: hex(o.successMostSubtle), Bold: new(false), }, }, @@ -228,7 +192,7 @@ func quickStyle(o quickStyleOpts) Styles { Bold: new(true), }, HorizontalRule: ansi.StylePrimitive{ - Color: hex(border), + Color: hex(o.separator), Format: "\n--------\n", }, Item: ansi.StylePrimitive{ @@ -247,7 +211,7 @@ func quickStyle(o quickStyleOpts) Styles { Underline: new(true), }, LinkText: ansi.StylePrimitive{ - Color: hex(successMuted), + Color: hex(o.successMostSubtle), Bold: new(true), }, Image: ansi.StylePrimitive{ @@ -255,40 +219,40 @@ func quickStyle(o quickStyleOpts) Styles { Underline: new(true), }, ImageText: ansi.StylePrimitive{ - Color: hex(fgMuted), + Color: hex(o.fgMoreSubtle), Format: "Image: {{.text}} →", }, Code: ansi.StyleBlock{ StylePrimitive: ansi.StylePrimitive{ Prefix: " ", Suffix: " ", - Color: hex(danger), - BackgroundColor: hex(bgSubtle), + Color: hex(o.destructive), + BackgroundColor: hex(o.bgLessVisible), }, }, CodeBlock: ansi.StyleCodeBlock{ StyleBlock: ansi.StyleBlock{ StylePrimitive: ansi.StylePrimitive{ - Color: hex(bgSubtle), + Color: hex(o.bgLessVisible), }, Margin: new(uint(defaultMargin)), }, Chroma: &ansi.Chroma{ Text: ansi.StylePrimitive{ - Color: hex(fgHalfMuted), + Color: hex(o.fgSubtle), }, Error: ansi.StylePrimitive{ - Color: hex(onAccent), - BackgroundColor: hex(error), + Color: hex(o.onPrimary), + BackgroundColor: hex(o.error), }, Comment: ansi.StylePrimitive{ - Color: hex(fgSubtle), + Color: hex(o.fgMostSubtle), }, CommentPreproc: ansi.StylePrimitive{ Color: hex(charmtone.Bengal), }, Keyword: ansi.StylePrimitive{ - Color: hex(info), + Color: hex(o.info), }, KeywordReserved: ansi.StylePrimitive{ Color: hex(charmtone.Pony), @@ -303,10 +267,10 @@ func quickStyle(o quickStyleOpts) Styles { Color: hex(charmtone.Salmon), }, Punctuation: ansi.StylePrimitive{ - Color: hex(warning), + Color: hex(o.warningSubtle), }, Name: ansi.StylePrimitive{ - Color: hex(fgHalfMuted), + Color: hex(o.fgSubtle), }, NameBuiltin: ansi.StylePrimitive{ Color: hex(charmtone.Cheeky), @@ -326,34 +290,34 @@ func quickStyle(o quickStyleOpts) Styles { Color: hex(charmtone.Citron), }, NameFunction: ansi.StylePrimitive{ - Color: hex(successMuted), + Color: hex(o.successMostSubtle), }, LiteralNumber: ansi.StylePrimitive{ - Color: hex(success), + Color: hex(o.success), }, LiteralString: ansi.StylePrimitive{ Color: hex(charmtone.Cumin), }, LiteralStringEscape: ansi.StylePrimitive{ - Color: hex(successSubtle), + Color: hex(o.successMoreSubtle), }, GenericDeleted: ansi.StylePrimitive{ - Color: hex(danger), + Color: hex(o.destructive), }, GenericEmph: ansi.StylePrimitive{ Italic: new(true), }, GenericInserted: ansi.StylePrimitive{ - Color: hex(successMuted), + Color: hex(o.successMostSubtle), }, GenericStrong: ansi.StylePrimitive{ Bold: new(true), }, GenericSubheading: ansi.StylePrimitive{ - Color: hex(fgMuted), + Color: hex(o.fgMoreSubtle), }, Background: ansi.StylePrimitive{ - BackgroundColor: hex(bgSubtle), + BackgroundColor: hex(o.bgLessVisible), }, }, }, @@ -368,8 +332,8 @@ func quickStyle(o quickStyleOpts) Styles { } // QuietMarkdown style - muted colors on subtle background for thinking content. - plainBg := hex(bgBaseLighter) - plainFg := hex(fgMuted) + plainBg := hex(o.bgLeastVisible) + plainFg := hex(o.fgMoreSubtle) s.QuietMarkdown = ansi.StyleConfig{ Document: ansi.StyleBlock{ StylePrimitive: ansi.StylePrimitive{ @@ -531,37 +495,37 @@ func quickStyle(o quickStyleOpts) Styles { } s.Help = help.Styles{ - ShortKey: base.Foreground(fgMuted), - ShortDesc: base.Foreground(fgSubtle), - ShortSeparator: base.Foreground(border), - Ellipsis: base.Foreground(border), - FullKey: base.Foreground(fgMuted), - FullDesc: base.Foreground(fgSubtle), - FullSeparator: base.Foreground(border), + ShortKey: base.Foreground(o.fgMoreSubtle), + ShortDesc: base.Foreground(o.fgMostSubtle), + ShortSeparator: base.Foreground(o.separator), + Ellipsis: base.Foreground(o.separator), + FullKey: base.Foreground(o.fgMoreSubtle), + FullDesc: base.Foreground(o.fgMostSubtle), + FullSeparator: base.Foreground(o.separator), } s.Diff = diffview.Style{ DividerLine: diffview.LineStyle{ LineNumber: lipgloss.NewStyle(). - Foreground(fgHalfMuted). - Background(bgBaseLighter), + Foreground(o.fgSubtle). + Background(o.bgLeastVisible), Code: lipgloss.NewStyle(). - Foreground(fgHalfMuted). - Background(bgBaseLighter), + Foreground(o.fgSubtle). + Background(o.bgLeastVisible), }, MissingLine: diffview.LineStyle{ LineNumber: lipgloss.NewStyle(). - Background(bgBaseLighter), + Background(o.bgLeastVisible), Code: lipgloss.NewStyle(). - Background(bgBaseLighter), + Background(o.bgLeastVisible), }, EqualLine: diffview.LineStyle{ LineNumber: lipgloss.NewStyle(). - Foreground(fgMuted). - Background(bgBase), + Foreground(o.fgMoreSubtle). + Background(o.bgBase), Code: lipgloss.NewStyle(). - Foreground(fgMuted). - Background(bgBase), + Foreground(o.fgMoreSubtle). + Background(o.bgBase), }, InsertLine: diffview.LineStyle{ LineNumber: lipgloss.NewStyle(). @@ -585,237 +549,252 @@ func quickStyle(o quickStyleOpts) Styles { }, Filename: diffview.LineStyle{ LineNumber: lipgloss.NewStyle(). - Foreground(fgHalfMuted). - Background(bgBaseLighter), + Foreground(o.fgSubtle). + Background(o.bgLeastVisible), Code: lipgloss.NewStyle(). - Foreground(fgHalfMuted). - Background(bgBaseLighter), + Foreground(o.fgSubtle). + Background(o.bgLeastVisible), }, } s.FilePicker = filepicker.Styles{ - DisabledCursor: base.Foreground(fgMuted), - Cursor: base.Foreground(fgBase), - Symlink: base.Foreground(fgSubtle), - Directory: base.Foreground(primary), - File: base.Foreground(fgBase), - DisabledFile: base.Foreground(fgMuted), - DisabledSelected: base.Background(bgOverlay).Foreground(fgMuted), - Permission: base.Foreground(fgMuted), - Selected: base.Background(primary).Foreground(fgBase), - FileSize: base.Foreground(fgMuted), - EmptyDirectory: base.Foreground(fgMuted).PaddingLeft(2).SetString("Empty directory"), + DisabledCursor: base.Foreground(o.fgMoreSubtle), + Cursor: base.Foreground(o.fgBase), + Symlink: base.Foreground(o.fgMostSubtle), + Directory: base.Foreground(o.primary), + File: base.Foreground(o.fgBase), + DisabledFile: base.Foreground(o.fgMoreSubtle), + DisabledSelected: base.Background(o.bgMostVisible).Foreground(o.fgMoreSubtle), + Permission: base.Foreground(o.fgMoreSubtle), + Selected: base.Background(o.primary).Foreground(o.fgBase), + FileSize: base.Foreground(o.fgMoreSubtle), + EmptyDirectory: base.Foreground(o.fgMoreSubtle).PaddingLeft(2).SetString("Empty directory"), } // borders - s.ToolCallSuccess = lipgloss.NewStyle().Foreground(success).SetString(ToolSuccess) + s.ToolCallSuccess = lipgloss.NewStyle().Foreground(o.success).SetString(ToolSuccess) - s.Header.Charm = base.Foreground(secondary) - s.Header.Diagonals = base.Foreground(primary) + s.Header.Charm = base.Foreground(o.secondary) + s.Header.Diagonals = base.Foreground(o.primary) s.Header.Percentage = muted + s.Header.Hypercredit = base.Foreground(charmtone.Dolly) s.Header.Keystroke = muted s.Header.KeystrokeTip = subtle s.Header.WorkingDir = muted s.Header.Separator = subtle - s.Header.Wrapper = lipgloss.NewStyle().Foreground(fgBase) + s.Header.Wrapper = lipgloss.NewStyle().Foreground(o.fgBase) s.Header.LogoGradCanvas = lipgloss.NewStyle() - s.Header.LogoGradFromColor = secondary - s.Header.LogoGradToColor = primary + s.Header.LogoGradFromColor = o.secondary + s.Header.LogoGradToColor = o.primary s.CompactDetails.Title = base - s.CompactDetails.View = base.Padding(0, 1, 1, 1).Border(lipgloss.RoundedBorder()).BorderForeground(borderFocus) - s.CompactDetails.Version = lipgloss.NewStyle().Foreground(border) + s.CompactDetails.View = base.Padding(0, 1, 1, 1).Border(lipgloss.RoundedBorder()).BorderForeground(o.primary) + s.CompactDetails.Version = lipgloss.NewStyle().Foreground(o.separator) // Tool rendering styles - s.Tool.IconPending = base.Foreground(successMuted).SetString(ToolPending) - s.Tool.IconSuccess = base.Foreground(success).SetString(ToolSuccess) - s.Tool.IconError = base.Foreground(error).SetString(ToolError) + s.Tool.IconPending = base.Foreground(o.successMostSubtle).SetString(ToolPending) + s.Tool.IconSuccess = base.Foreground(o.success).SetString(ToolSuccess) + s.Tool.IconError = base.Foreground(o.error).SetString(ToolError) s.Tool.IconCancelled = muted.SetString(ToolPending) - s.Tool.NameNormal = base.Foreground(info) - s.Tool.NameNested = base.Foreground(info) + s.Tool.NameNormal = base.Foreground(o.info) + s.Tool.NameNested = base.Foreground(o.info) s.Tool.ParamMain = subtle s.Tool.ParamKey = subtle // Content rendering - prepared styles that accept width parameter - s.Tool.ContentLine = muted.Background(bgBaseLighter) - s.Tool.ContentTruncation = muted.Background(bgBaseLighter) - s.Tool.ContentCodeLine = base.Background(bgBase).PaddingLeft(2) - s.Tool.ContentCodeTruncation = muted.Background(bgBase).PaddingLeft(2) - s.Tool.ContentCodeBg = bgBase + s.Tool.ContentLine = muted.Background(o.bgLeastVisible) + s.Tool.ContentTruncation = muted.Background(o.bgLeastVisible) + s.Tool.ContentCodeLine = base.Background(o.bgBase).PaddingLeft(2) + s.Tool.ContentCodeTruncation = muted.Background(o.bgBase).PaddingLeft(2) + s.Tool.ContentCodeBg = o.bgBase s.Tool.Body = base.PaddingLeft(2) // Deprecated - kept for backward compatibility - s.Tool.ContentBg = muted.Background(bgBaseLighter) + s.Tool.ContentBg = muted.Background(o.bgLeastVisible) s.Tool.ContentText = muted - s.Tool.ContentLineNumber = base.Foreground(fgMuted).Background(bgBase).PaddingRight(1).PaddingLeft(1) + s.Tool.ContentLineNumber = base.Foreground(o.fgMoreSubtle).Background(o.bgBase).PaddingRight(1).PaddingLeft(1) - s.Tool.StateWaiting = base.Foreground(fgSubtle) - s.Tool.StateCancelled = base.Foreground(fgSubtle) + s.Tool.StateWaiting = base.Foreground(o.fgMostSubtle) + s.Tool.StateCancelled = base.Foreground(o.fgMostSubtle) - s.Tool.ErrorTag = base.Padding(0, 1).Background(danger).Foreground(onAccent) - s.Tool.ErrorMessage = base.Foreground(fgHalfMuted) + s.Tool.ErrorTag = base.Padding(0, 1).Background(o.destructive).Foreground(o.onPrimary) + s.Tool.ErrorMessage = base.Foreground(o.fgSubtle) // Diff and multi-edit styles - s.Tool.DiffTruncation = muted.Background(bgBaseLighter).PaddingLeft(2) - s.Tool.NoteTag = base.Padding(0, 1).Background(info).Foreground(onAccent) - s.Tool.NoteMessage = base.Foreground(fgHalfMuted) + s.Tool.DiffTruncation = muted.Background(o.bgLeastVisible).PaddingLeft(2) + s.Tool.NoteTag = base.Padding(0, 1).Background(o.info).Foreground(o.onPrimary) + s.Tool.NoteMessage = base.Foreground(o.fgSubtle) // Job header styles - s.Tool.JobIconPending = base.Foreground(successMuted) - s.Tool.JobIconError = base.Foreground(error) - s.Tool.JobIconSuccess = base.Foreground(success) - s.Tool.JobToolName = base.Foreground(info) - s.Tool.JobAction = base.Foreground(infoMuted) + s.Tool.JobIconPending = base.Foreground(o.successMostSubtle) + s.Tool.JobIconError = base.Foreground(o.error) + s.Tool.JobIconSuccess = base.Foreground(o.success) + s.Tool.JobToolName = base.Foreground(o.info) + s.Tool.JobAction = base.Foreground(o.infoMostSubtle) s.Tool.JobPID = muted s.Tool.JobDescription = subtle // Agent task styles - s.Tool.AgentTaskTag = base.Bold(true).Padding(0, 1).MarginLeft(2).Background(infoSubtle).Foreground(onAccent) + s.Tool.AgentTaskTag = base.Bold(true).Padding(0, 1).MarginLeft(2).Background(o.infoMoreSubtle).Foreground(o.onPrimary) s.Tool.AgentPrompt = muted // Agentic fetch styles - s.Tool.AgenticFetchPromptTag = base.Bold(true).Padding(0, 1).MarginLeft(2).Background(success).Foreground(border) + s.Tool.AgenticFetchPromptTag = base.Bold(true).Padding(0, 1).MarginLeft(2).Background(o.success).Foreground(o.separator) // Todo styles - s.Tool.TodoRatio = base.Foreground(infoMuted) - s.Tool.TodoCompletedIcon = base.Foreground(success) - s.Tool.TodoInProgressIcon = base.Foreground(successMuted) - s.Tool.TodoPendingIcon = base.Foreground(fgMuted) - s.Tool.TodoStatusNote = lipgloss.NewStyle().Foreground(fgSubtle) - s.Tool.TodoItem = lipgloss.NewStyle().Foreground(fgBase) - s.Tool.TodoJustStarted = lipgloss.NewStyle().Foreground(fgBase) + s.Tool.TodoRatio = base.Foreground(o.infoMostSubtle) + s.Tool.TodoCompletedIcon = base.Foreground(o.success) + s.Tool.TodoInProgressIcon = base.Foreground(o.successMostSubtle) + s.Tool.TodoPendingIcon = base.Foreground(o.fgMoreSubtle) + s.Tool.TodoStatusNote = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Tool.TodoItem = lipgloss.NewStyle().Foreground(o.fgBase) + s.Tool.TodoJustStarted = lipgloss.NewStyle().Foreground(o.fgBase) // MCP styles - s.Tool.MCPName = base.Foreground(info) - s.Tool.MCPToolName = base.Foreground(infoMuted) - s.Tool.MCPArrow = base.Foreground(info).SetString(ArrowRightIcon) + s.Tool.MCPName = base.Foreground(o.info) + s.Tool.MCPToolName = base.Foreground(o.infoMostSubtle) + s.Tool.MCPArrow = base.Foreground(o.info).SetString(ArrowRightIcon) // Loading indicators for images, skills - s.Tool.ResourceLoadedText = base.Foreground(success) - s.Tool.ResourceLoadedIndicator = base.Foreground(successMuted) + s.Tool.ResourceLoadedText = base.Foreground(o.success) + s.Tool.ResourceLoadedIndicator = base.Foreground(o.successMostSubtle) s.Tool.ResourceName = base s.Tool.MediaType = base - s.Tool.ResourceSize = base.Foreground(fgMuted) + s.Tool.ResourceSize = base.Foreground(o.fgMoreSubtle) + + // Hook styles + s.Tool.HookLabel = base.Foreground(o.successMoreSubtle) + s.Tool.HookName = base + s.Tool.HookMatcher = base.Foreground(o.fgMoreSubtle) + s.Tool.HookArrow = base.Foreground(o.successMoreSubtle) + s.Tool.HookDetail = base.Foreground(o.fgMoreSubtle) + s.Tool.HookOK = base.Foreground(o.successMostSubtle) + s.Tool.HookDenied = base.Foreground(o.error) + s.Tool.HookDeniedLabel = base.Foreground(o.destructive) + s.Tool.HookDeniedReason = base.Foreground(o.bgMostVisible) + s.Tool.HookRewrote = base.Foreground(o.bgMostVisible) // Tool-call action verbs and result-list styling. - s.Tool.ActionCreate = lipgloss.NewStyle().Foreground(successSubtle) - s.Tool.ActionDestroy = lipgloss.NewStyle().Foreground(danger) - s.Tool.ResultEmpty = lipgloss.NewStyle().Foreground(fgSubtle) - s.Tool.ResultTruncation = lipgloss.NewStyle().Foreground(fgSubtle) - s.Tool.ResultItemName = lipgloss.NewStyle().Foreground(fgBase) - s.Tool.ResultItemDesc = lipgloss.NewStyle().Foreground(fgSubtle) + s.Tool.ActionCreate = lipgloss.NewStyle().Foreground(o.successMoreSubtle) + s.Tool.ActionDestroy = lipgloss.NewStyle().Foreground(o.destructive) + s.Tool.ResultEmpty = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Tool.ResultTruncation = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Tool.ResultItemName = lipgloss.NewStyle().Foreground(o.fgBase) + s.Tool.ResultItemDesc = lipgloss.NewStyle().Foreground(o.fgMostSubtle) // Buttons - s.Button.Focused = lipgloss.NewStyle().Foreground(onAccent).Background(secondary) - s.Button.Blurred = lipgloss.NewStyle().Foreground(fgBase).Background(bgSubtle) + s.Button.Focused = lipgloss.NewStyle().Foreground(o.onPrimary).Background(o.secondary) + s.Button.Blurred = lipgloss.NewStyle().Foreground(o.fgBase).Background(o.bgLessVisible) // Editor - s.Editor.PromptNormalFocused = lipgloss.NewStyle().Foreground(successMuted).SetString("::: ") - s.Editor.PromptNormalBlurred = s.Editor.PromptNormalFocused.Foreground(fgMuted) - s.Editor.PromptYoloIconFocused = lipgloss.NewStyle().MarginRight(1).Foreground(fgSubtle).Background(busy).Bold(true).SetString(" ! ") - s.Editor.PromptYoloIconBlurred = s.Editor.PromptYoloIconFocused.Foreground(bgBase).Background(fgMuted) - s.Editor.PromptYoloDotsFocused = lipgloss.NewStyle().MarginRight(1).Foreground(warning).SetString(":::") - s.Editor.PromptYoloDotsBlurred = s.Editor.PromptYoloDotsFocused.Foreground(fgMuted) + s.Editor.PromptNormalFocused = lipgloss.NewStyle().Foreground(o.successMostSubtle).SetString("::: ") + s.Editor.PromptNormalBlurred = s.Editor.PromptNormalFocused.Foreground(o.fgMoreSubtle) + s.Editor.PromptYoloIconFocused = lipgloss.NewStyle().MarginRight(1).Foreground(o.fgMostSubtle).Background(o.busy).Bold(true).SetString(" ! ") + s.Editor.PromptYoloIconBlurred = s.Editor.PromptYoloIconFocused.Foreground(o.bgBase).Background(o.fgMoreSubtle) + s.Editor.PromptYoloDotsFocused = lipgloss.NewStyle().MarginRight(1).Foreground(o.warningSubtle).SetString(":::") + s.Editor.PromptYoloDotsBlurred = s.Editor.PromptYoloDotsFocused.Foreground(o.fgMoreSubtle) - s.Radio.On = lipgloss.NewStyle().Foreground(fgHalfMuted).SetString(RadioOn) - s.Radio.Off = lipgloss.NewStyle().Foreground(fgHalfMuted).SetString(RadioOff) - s.Radio.Label = lipgloss.NewStyle().Foreground(fgHalfMuted) + s.Radio.On = lipgloss.NewStyle().Foreground(o.fgSubtle).SetString(RadioOn) + s.Radio.Off = lipgloss.NewStyle().Foreground(o.fgSubtle).SetString(RadioOff) + s.Radio.Label = lipgloss.NewStyle().Foreground(o.fgSubtle) // Logo - s.Logo.FieldColor = primary - s.Logo.TitleColorA = secondary - s.Logo.TitleColorB = primary - s.Logo.CharmColor = secondary - s.Logo.VersionColor = primary - s.Logo.SmallCharm = lipgloss.NewStyle().Foreground(secondary) - s.Logo.SmallDiagonals = lipgloss.NewStyle().Foreground(primary) + s.Logo.FieldColor = o.primary + s.Logo.TitleColorA = o.secondary + s.Logo.TitleColorB = o.primary + s.Logo.CharmColor = o.secondary + s.Logo.VersionColor = o.primary + s.Logo.SmallCharm = lipgloss.NewStyle().Foreground(o.secondary) + s.Logo.SmallDiagonals = lipgloss.NewStyle().Foreground(o.primary) s.Logo.GradCanvas = lipgloss.NewStyle() - s.Logo.SmallGradFromColor = secondary - s.Logo.SmallGradToColor = primary + s.Logo.SmallGradFromColor = o.secondary + s.Logo.SmallGradToColor = o.primary // Section s.Section.Title = subtle - s.Section.Line = base.Foreground(border) + s.Section.Line = base.Foreground(o.separator) // Initialize s.Initialize.Header = base s.Initialize.Content = muted - s.Initialize.Accent = base.Foreground(successMuted) + s.Initialize.Accent = base.Foreground(o.successMostSubtle) // ResourceGroup (LSP/MCP/skills sidebar lists). - s.Resource.Heading = lipgloss.NewStyle().Foreground(fgSubtle) - s.Resource.Name = lipgloss.NewStyle().Foreground(fgMuted) - s.Resource.StatusText = lipgloss.NewStyle().Foreground(fgSubtle) - s.Resource.OfflineIcon = lipgloss.NewStyle().Foreground(bgOverlay).SetString("●") - s.Resource.BusyIcon = s.Resource.OfflineIcon.Foreground(busy) - s.Resource.ErrorIcon = s.Resource.OfflineIcon.Foreground(danger) - s.Resource.OnlineIcon = s.Resource.OfflineIcon.Foreground(successMuted) - s.Resource.DisabledIcon = lipgloss.NewStyle().Foreground(fgMuted).SetString("●") - s.Resource.AdditionalText = lipgloss.NewStyle().Foreground(fgSubtle) - s.Resource.CapabilityCount = lipgloss.NewStyle().Foreground(fgSubtle) - s.Resource.RowTitleBase = lipgloss.NewStyle().Foreground(fgBase) - s.Resource.RowDescBase = lipgloss.NewStyle().Foreground(fgBase) - s.Resource.DefaultTitleFg = fgMuted - s.Resource.DefaultDescFg = fgSubtle + s.Resource.Heading = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Resource.Name = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Resource.StatusText = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Resource.OfflineIcon = lipgloss.NewStyle().Foreground(o.bgMostVisible).SetString("●") + s.Resource.BusyIcon = s.Resource.OfflineIcon.Foreground(o.busy) + s.Resource.ErrorIcon = s.Resource.OfflineIcon.Foreground(o.destructive) + s.Resource.OnlineIcon = s.Resource.OfflineIcon.Foreground(o.successMostSubtle) + s.Resource.DisabledIcon = lipgloss.NewStyle().Foreground(o.fgMoreSubtle).SetString("●") + s.Resource.AdditionalText = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Resource.CapabilityCount = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Resource.RowTitleBase = lipgloss.NewStyle().Foreground(o.fgBase) + s.Resource.RowDescBase = lipgloss.NewStyle().Foreground(o.fgBase) + s.Resource.DefaultTitleFg = o.fgMoreSubtle + s.Resource.DefaultDescFg = o.fgMostSubtle // LSP - s.LSP.ErrorDiagnostic = base.Foreground(error) - s.LSP.WarningDiagnostic = base.Foreground(warning) - s.LSP.HintDiagnostic = base.Foreground(fgHalfMuted) - s.LSP.InfoDiagnostic = base.Foreground(info) + s.LSP.ErrorDiagnostic = base.Foreground(o.error) + s.LSP.WarningDiagnostic = base.Foreground(o.warningSubtle) + s.LSP.HintDiagnostic = base.Foreground(o.fgSubtle) + s.LSP.InfoDiagnostic = base.Foreground(o.info) // Files - s.Files.Path = lipgloss.NewStyle().Foreground(fgMuted) - s.Files.Additions = lipgloss.NewStyle().Foreground(successMuted) - s.Files.Deletions = lipgloss.NewStyle().Foreground(error) - s.Files.SectionTitle = lipgloss.NewStyle().Foreground(fgSubtle) - s.Files.EmptyMessage = lipgloss.NewStyle().Foreground(fgSubtle) - s.Files.TruncationHint = lipgloss.NewStyle().Foreground(fgSubtle) + s.Files.Path = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Files.Additions = lipgloss.NewStyle().Foreground(o.successMostSubtle) + s.Files.Deletions = lipgloss.NewStyle().Foreground(o.error) + s.Files.SectionTitle = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Files.EmptyMessage = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Files.TruncationHint = lipgloss.NewStyle().Foreground(o.fgMostSubtle) // Sidebar - s.Sidebar.SessionTitle = lipgloss.NewStyle().Foreground(fgMuted) - s.Sidebar.WorkingDir = lipgloss.NewStyle().Foreground(fgMuted) + s.Sidebar.SessionTitle = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Sidebar.WorkingDir = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) // ModelInfo - s.ModelInfo.Icon = lipgloss.NewStyle().Foreground(fgSubtle) - s.ModelInfo.Name = lipgloss.NewStyle().Foreground(fgBase) - s.ModelInfo.Provider = lipgloss.NewStyle().Foreground(fgMuted) - s.ModelInfo.ProviderFallback = lipgloss.NewStyle().Foreground(fgMuted).PaddingLeft(2) - s.ModelInfo.Reasoning = lipgloss.NewStyle().Foreground(fgSubtle).PaddingLeft(2) - s.ModelInfo.TokenCount = lipgloss.NewStyle().Foreground(fgSubtle) - s.ModelInfo.TokenPercentage = lipgloss.NewStyle().Foreground(fgMuted) - s.ModelInfo.Cost = lipgloss.NewStyle().Foreground(fgMuted) + s.ModelInfo.Icon = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.ModelInfo.Name = lipgloss.NewStyle().Foreground(o.fgBase) + s.ModelInfo.Provider = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.ModelInfo.ProviderFallback = lipgloss.NewStyle().Foreground(o.fgMoreSubtle).PaddingLeft(2) + s.ModelInfo.Reasoning = lipgloss.NewStyle().Foreground(o.fgMostSubtle).PaddingLeft(2) + s.ModelInfo.TokenCount = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.ModelInfo.TokenPercentage = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.ModelInfo.Cost = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.ModelInfo.HypercreditIcon = lipgloss.NewStyle().Foreground(charmtone.Dolly) + s.ModelInfo.HypercreditText = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) // ResourceGroup - s.Resource.DefaultTitleFg = fgMuted - s.Resource.DefaultDescFg = fgSubtle + s.Resource.DefaultTitleFg = o.fgMoreSubtle + s.Resource.DefaultDescFg = o.fgMostSubtle // Chat messageFocussedBorder := lipgloss.Border{ Left: "▌", } - s.Messages.NoContent = lipgloss.NewStyle().Foreground(fgBase) + s.Messages.NoContent = lipgloss.NewStyle().Foreground(o.fgBase) s.Messages.UserBlurred = s.Messages.NoContent.PaddingLeft(1).BorderLeft(true). - BorderForeground(primary).BorderStyle(lipgloss.NormalBorder()) + BorderForeground(o.primary).BorderStyle(lipgloss.NormalBorder()) s.Messages.UserFocused = s.Messages.NoContent.PaddingLeft(1).BorderLeft(true). - BorderForeground(primary).BorderStyle(messageFocussedBorder) + BorderForeground(o.primary).BorderStyle(messageFocussedBorder) s.Messages.AssistantBlurred = s.Messages.NoContent.PaddingLeft(2) s.Messages.AssistantFocused = s.Messages.NoContent.PaddingLeft(1).BorderLeft(true). - BorderForeground(successMuted).BorderStyle(messageFocussedBorder) + BorderForeground(o.successMostSubtle).BorderStyle(messageFocussedBorder) s.Messages.Thinking = lipgloss.NewStyle().MaxHeight(10) s.Messages.ErrorTag = lipgloss.NewStyle().Padding(0, 1). - Background(danger).Foreground(onAccent) - s.Messages.ErrorTitle = lipgloss.NewStyle().Foreground(fgHalfMuted) - s.Messages.ErrorDetails = lipgloss.NewStyle().Foreground(fgSubtle) + Background(o.destructive).Foreground(o.onPrimary) + s.Messages.ErrorTitle = lipgloss.NewStyle().Foreground(o.fgSubtle) + s.Messages.ErrorDetails = lipgloss.NewStyle().Foreground(o.fgMostSubtle) // Message item styles s.Messages.ToolCallFocused = muted.PaddingLeft(1). BorderStyle(messageFocussedBorder). BorderLeft(true). - BorderForeground(successMuted) + BorderForeground(o.successMostSubtle) s.Messages.ToolCallBlurred = muted.PaddingLeft(2) // No padding or border for compact tool calls within messages s.Messages.ToolCallCompact = muted @@ -824,144 +803,144 @@ func quickStyle(o quickStyleOpts) Styles { s.Messages.AssistantInfoModel = muted s.Messages.AssistantInfoProvider = subtle s.Messages.AssistantInfoDuration = subtle - s.Messages.AssistantCanceled = lipgloss.NewStyle().Foreground(fgBase).Italic(true) + s.Messages.AssistantCanceled = lipgloss.NewStyle().Foreground(o.fgBase).Italic(true) // Thinking section styles - s.Messages.ThinkingBox = subtle.Background(bgBaseLighter) + s.Messages.ThinkingBox = subtle.Background(o.bgLeastVisible) s.Messages.ThinkingTruncationHint = muted s.Messages.ThinkingFooterTitle = muted s.Messages.ThinkingFooterDuration = subtle // Text selection. - s.TextSelection = lipgloss.NewStyle().Foreground(onPrimary).Background(primary) + s.TextSelection = lipgloss.NewStyle().Foreground(o.onPrimary).Background(o.primary) // Dialog styles - s.Dialog.Title = base.Padding(0, 1).Foreground(primary) - s.Dialog.TitleText = base.Foreground(primary) - s.Dialog.TitleError = base.Foreground(danger) - s.Dialog.TitleAccent = base.Foreground(success).Bold(true) + s.Dialog.Title = base.Padding(0, 1).Foreground(o.primary) + s.Dialog.TitleText = base.Foreground(o.primary) + s.Dialog.TitleError = base.Foreground(o.destructive) + s.Dialog.TitleAccent = base.Foreground(o.success).Bold(true) s.Dialog.TitleLineBase = lipgloss.NewStyle() - s.Dialog.TitleGradFromColor = primary - s.Dialog.TitleGradToColor = secondary + s.Dialog.TitleGradFromColor = o.primary + s.Dialog.TitleGradToColor = o.secondary // Dialog.ListItem (commands, reasoning, models) - s.Dialog.ListItem.InfoBlurred = lipgloss.NewStyle().Foreground(fgBase) - s.Dialog.ListItem.InfoFocused = lipgloss.NewStyle().Foreground(fgBase) + s.Dialog.ListItem.InfoBlurred = lipgloss.NewStyle().Foreground(o.fgBase) + s.Dialog.ListItem.InfoFocused = lipgloss.NewStyle().Foreground(o.fgBase) // Dialog.Models - s.Dialog.Models.ConfiguredText = lipgloss.NewStyle().Foreground(fgSubtle) + s.Dialog.Models.ConfiguredText = lipgloss.NewStyle().Foreground(o.fgMostSubtle) // Dialog.Permissions - s.Dialog.Permissions.KeyText = lipgloss.NewStyle().Foreground(fgMuted) - s.Dialog.Permissions.ValueText = lipgloss.NewStyle().Foreground(fgBase) - s.Dialog.Permissions.ParamsBg = bgSubtle + s.Dialog.Permissions.KeyText = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Dialog.Permissions.ValueText = lipgloss.NewStyle().Foreground(o.fgBase) + s.Dialog.Permissions.ParamsBg = o.bgLessVisible // Dialog.Quit - s.Dialog.Quit.Content = lipgloss.NewStyle().Foreground(fgBase) - s.Dialog.Quit.Frame = lipgloss.NewStyle().BorderForeground(borderFocus).Border(lipgloss.RoundedBorder()).Padding(1, 2) - s.Dialog.View = base.Border(lipgloss.RoundedBorder()).BorderForeground(borderFocus) - s.Dialog.PrimaryText = base.Padding(0, 1).Foreground(primary) - s.Dialog.SecondaryText = base.Padding(0, 1).Foreground(fgSubtle) + s.Dialog.Quit.Content = lipgloss.NewStyle().Foreground(o.fgBase) + s.Dialog.Quit.Frame = lipgloss.NewStyle().BorderForeground(o.primary).Border(lipgloss.RoundedBorder()).Padding(1, 2) + s.Dialog.View = base.Border(lipgloss.RoundedBorder()).BorderForeground(o.primary) + s.Dialog.PrimaryText = base.Padding(0, 1).Foreground(o.primary) + s.Dialog.SecondaryText = base.Padding(0, 1).Foreground(o.fgMostSubtle) s.Dialog.HelpView = base.Padding(0, 1).AlignHorizontal(lipgloss.Left) - s.Dialog.Help.ShortKey = base.Foreground(fgMuted) - s.Dialog.Help.ShortDesc = base.Foreground(fgSubtle) - s.Dialog.Help.ShortSeparator = base.Foreground(border) - s.Dialog.Help.Ellipsis = base.Foreground(border) - s.Dialog.Help.FullKey = base.Foreground(fgMuted) - s.Dialog.Help.FullDesc = base.Foreground(fgSubtle) - s.Dialog.Help.FullSeparator = base.Foreground(border) - s.Dialog.NormalItem = base.Padding(0, 1).Foreground(fgBase) - s.Dialog.SelectedItem = base.Padding(0, 1).Background(primary).Foreground(fgBase) + s.Dialog.Help.ShortKey = base.Foreground(o.fgMoreSubtle) + s.Dialog.Help.ShortDesc = base.Foreground(o.fgMostSubtle) + s.Dialog.Help.ShortSeparator = base.Foreground(o.separator) + s.Dialog.Help.Ellipsis = base.Foreground(o.separator) + s.Dialog.Help.FullKey = base.Foreground(o.fgMoreSubtle) + s.Dialog.Help.FullDesc = base.Foreground(o.fgMostSubtle) + s.Dialog.Help.FullSeparator = base.Foreground(o.separator) + s.Dialog.NormalItem = base.Padding(0, 1).Foreground(o.fgBase) + s.Dialog.SelectedItem = base.Padding(0, 1).Background(o.primary).Foreground(o.onPrimary) s.Dialog.InputPrompt = base.Margin(1, 1) s.Dialog.List = base.Margin(0, 0, 1, 0) - s.Dialog.ContentPanel = base.Background(bgSubtle).Foreground(fgBase).Padding(1, 2) - s.Dialog.Spinner = base.Foreground(secondary) - s.Dialog.ScrollbarThumb = base.Foreground(secondary) - s.Dialog.ScrollbarTrack = base.Foreground(border) + s.Dialog.ContentPanel = base.Background(o.bgLessVisible).Foreground(o.fgBase).Padding(1, 2) + s.Dialog.Spinner = base.Foreground(o.secondary) + s.Dialog.ScrollbarThumb = base.Foreground(o.secondary) + s.Dialog.ScrollbarTrack = base.Foreground(o.separator) - s.Dialog.ImagePreview = lipgloss.NewStyle().Padding(0, 1).Foreground(fgSubtle) + s.Dialog.ImagePreview = lipgloss.NewStyle().Padding(0, 1).Foreground(o.fgMostSubtle) // API key input dialog - s.Dialog.APIKey.Spinner = base.Foreground(success) + s.Dialog.APIKey.Spinner = base.Foreground(o.success) // OAuth dialog - s.Dialog.OAuth.Spinner = base.Foreground(successSubtle) - s.Dialog.OAuth.Instructions = lipgloss.NewStyle().Foreground(onAccent) - s.Dialog.OAuth.UserCode = lipgloss.NewStyle().Bold(true).Foreground(onAccent) - s.Dialog.OAuth.Success = lipgloss.NewStyle().Foreground(successSubtle) - s.Dialog.OAuth.Link = lipgloss.NewStyle().Foreground(successMuted).Underline(true) - s.Dialog.OAuth.Enter = lipgloss.NewStyle().Foreground(primary) - s.Dialog.OAuth.ErrorText = lipgloss.NewStyle().Foreground(error) - s.Dialog.OAuth.StatusText = lipgloss.NewStyle().Foreground(fgMuted) - s.Dialog.OAuth.UserCodeBg = bgBaseLighter + s.Dialog.OAuth.Spinner = base.Foreground(o.successMoreSubtle) + s.Dialog.OAuth.Instructions = lipgloss.NewStyle().Foreground(o.fgBase) + s.Dialog.OAuth.UserCode = lipgloss.NewStyle().Bold(true).Foreground(o.fgBase) + s.Dialog.OAuth.Success = lipgloss.NewStyle().Foreground(o.successMoreSubtle) + s.Dialog.OAuth.Link = lipgloss.NewStyle().Foreground(o.successMostSubtle).Underline(true) + s.Dialog.OAuth.Enter = lipgloss.NewStyle().Foreground(o.keyword) + s.Dialog.OAuth.ErrorText = lipgloss.NewStyle().Foreground(o.error) + s.Dialog.OAuth.StatusText = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Dialog.OAuth.UserCodeBg = o.bgLeastVisible s.Dialog.Arguments.Content = base.Padding(1) s.Dialog.Arguments.Description = base.MarginBottom(1).MaxHeight(3) - s.Dialog.Arguments.InputLabelBlurred = base.Foreground(fgMuted) + s.Dialog.Arguments.InputLabelBlurred = base.Foreground(o.fgMoreSubtle) s.Dialog.Arguments.InputLabelFocused = base.Bold(true) - s.Dialog.Arguments.InputRequiredMarkBlurred = base.Foreground(fgMuted).SetString("*") - s.Dialog.Arguments.InputRequiredMarkFocused = base.Foreground(primary).Bold(true).SetString("*") + s.Dialog.Arguments.InputRequiredMarkBlurred = base.Foreground(o.fgMoreSubtle).SetString("*") + s.Dialog.Arguments.InputRequiredMarkFocused = base.Foreground(o.primary).Bold(true).SetString("*") - s.Dialog.Sessions.DeletingTitle = s.Dialog.Title.Foreground(danger) - s.Dialog.Sessions.DeletingView = s.Dialog.View.BorderForeground(danger) + s.Dialog.Sessions.DeletingTitle = s.Dialog.Title.Foreground(o.destructive) + s.Dialog.Sessions.DeletingView = s.Dialog.View.BorderForeground(o.destructive) s.Dialog.Sessions.DeletingMessage = base.Padding(1) - s.Dialog.Sessions.DeletingTitleGradientFromColor = danger - s.Dialog.Sessions.DeletingTitleGradientToColor = primary - s.Dialog.Sessions.DeletingItemBlurred = s.Dialog.NormalItem.Foreground(fgSubtle) - s.Dialog.Sessions.DeletingItemFocused = s.Dialog.SelectedItem.Background(danger).Foreground(onAccent) + s.Dialog.Sessions.DeletingTitleGradientFromColor = o.destructive + s.Dialog.Sessions.DeletingTitleGradientToColor = o.primary + s.Dialog.Sessions.DeletingItemBlurred = s.Dialog.NormalItem.Foreground(o.fgMostSubtle) + s.Dialog.Sessions.DeletingItemFocused = s.Dialog.SelectedItem.Background(o.destructive).Foreground(o.onPrimary) - s.Dialog.Sessions.RenamingingTitle = s.Dialog.Title.Foreground(warning) - s.Dialog.Sessions.RenamingView = s.Dialog.View.BorderForeground(warning) + s.Dialog.Sessions.RenamingingTitle = s.Dialog.Title.Foreground(o.warningSubtle) + s.Dialog.Sessions.RenamingView = s.Dialog.View.BorderForeground(o.warningSubtle) s.Dialog.Sessions.RenamingingMessage = base.Padding(1) - s.Dialog.Sessions.RenamingTitleGradientFromColor = warning - s.Dialog.Sessions.RenamingTitleGradientToColor = tertiary - s.Dialog.Sessions.RenamingItemBlurred = s.Dialog.NormalItem.Foreground(fgSubtle) + s.Dialog.Sessions.RenamingTitleGradientFromColor = o.warningSubtle + s.Dialog.Sessions.RenamingTitleGradientToColor = o.accent + s.Dialog.Sessions.RenamingItemBlurred = s.Dialog.NormalItem.Foreground(o.fgMostSubtle) s.Dialog.Sessions.RenamingingItemFocused = s.Dialog.SelectedItem.UnsetBackground().UnsetForeground() - s.Dialog.Sessions.RenamingPlaceholder = base.Foreground(fgMuted) - s.Dialog.Sessions.InfoBlurred = lipgloss.NewStyle().Foreground(fgSubtle) - s.Dialog.Sessions.InfoFocused = lipgloss.NewStyle().Foreground(fgBase) + s.Dialog.Sessions.RenamingPlaceholder = base.Foreground(o.fgMoreSubtle) + s.Dialog.Sessions.InfoBlurred = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Dialog.Sessions.InfoFocused = lipgloss.NewStyle().Foreground(o.fgBase) s.Status.Help = lipgloss.NewStyle().Padding(0, 1) - s.Status.SuccessIndicator = base.Foreground(bgSubtle).Background(success).Padding(0, 1).Bold(true).SetString("OKAY!") + s.Status.SuccessIndicator = base.Foreground(o.bgLessVisible).Background(o.success).Padding(0, 1).Bold(true).SetString("OKAY!") s.Status.InfoIndicator = s.Status.SuccessIndicator s.Status.UpdateIndicator = s.Status.SuccessIndicator.SetString("HEY!") - s.Status.WarnIndicator = s.Status.SuccessIndicator.Foreground(bgOverlay).Background(warningStrong).SetString("WARNING") - s.Status.ErrorIndicator = s.Status.SuccessIndicator.Foreground(bgBase).Background(danger).SetString("ERROR") - s.Status.SuccessMessage = base.Foreground(bgSubtle).Background(successMuted).Padding(0, 1) + s.Status.WarnIndicator = s.Status.SuccessIndicator.Foreground(o.bgMostVisible).Background(o.warning).SetString("WARNING") + s.Status.ErrorIndicator = s.Status.SuccessIndicator.Foreground(o.bgBase).Background(o.destructive).SetString("ERROR") + s.Status.SuccessMessage = base.Foreground(o.bgLessVisible).Background(o.successMostSubtle).Padding(0, 1) s.Status.InfoMessage = s.Status.SuccessMessage s.Status.UpdateMessage = s.Status.SuccessMessage - s.Status.WarnMessage = s.Status.SuccessMessage.Foreground(bgOverlay).Background(warning) - s.Status.ErrorMessage = s.Status.SuccessMessage.Foreground(onAccent).Background(error) + s.Status.WarnMessage = s.Status.SuccessMessage.Foreground(o.bgMostVisible).Background(o.warningSubtle) + s.Status.ErrorMessage = s.Status.SuccessMessage.Foreground(o.onPrimary).Background(o.error) // Completions styles - s.Completions.Normal = base.Background(bgSubtle).Foreground(fgBase) - s.Completions.Focused = base.Background(primary).Foreground(onAccent) + s.Completions.Normal = base.Background(o.bgLessVisible).Foreground(o.fgBase) + s.Completions.Focused = base.Background(o.primary).Foreground(o.onPrimary) s.Completions.Match = base.Underline(true) // Attachments styles - attachmentIconStyle := base.Foreground(bgSubtle).Background(success).Padding(0, 1) + attachmentIconStyle := base.Foreground(o.bgLessVisible).Background(o.success).Padding(0, 1) s.Attachments.Image = attachmentIconStyle.SetString(ImageIcon) s.Attachments.Text = attachmentIconStyle.SetString(TextIcon) - s.Attachments.Normal = base.Padding(0, 1).MarginRight(1).Background(fgMuted).Foreground(fgBase) - s.Attachments.Deleting = base.Padding(0, 1).Bold(true).Background(danger).Foreground(fgBase) + s.Attachments.Normal = base.Padding(0, 1).MarginRight(1).Background(o.fgMoreSubtle).Foreground(o.fgBase) + s.Attachments.Deleting = base.Padding(0, 1).Bold(true).Background(o.destructive).Foreground(o.fgBase) // Pills styles s.Pills.Base = base.Padding(0, 1) - s.Pills.Focused = base.Padding(0, 1).BorderStyle(lipgloss.RoundedBorder()).BorderForeground(bgOverlay) + s.Pills.Focused = base.Padding(0, 1).BorderStyle(lipgloss.RoundedBorder()).BorderForeground(o.bgMostVisible) s.Pills.Blurred = base.Padding(0, 1).BorderStyle(lipgloss.HiddenBorder()) - s.Pills.QueueItemPrefix = lipgloss.NewStyle().Foreground(fgMuted).SetString(" •") - s.Pills.QueueItemText = lipgloss.NewStyle().Foreground(fgMuted) - s.Pills.QueueLabel = lipgloss.NewStyle().Foreground(fgBase) - s.Pills.QueueIconBase = lipgloss.NewStyle().Foreground(fgBase) - s.Pills.QueueGradFromColor = error - s.Pills.QueueGradToColor = secondary - s.Pills.TodoLabel = lipgloss.NewStyle().Foreground(fgBase) - s.Pills.TodoProgress = lipgloss.NewStyle().Foreground(fgMuted) - s.Pills.TodoCurrentTask = lipgloss.NewStyle().Foreground(fgSubtle) - s.Pills.TodoSpinner = lipgloss.NewStyle().Foreground(successMuted) - s.Pills.HelpKey = lipgloss.NewStyle().Foreground(fgMuted) - s.Pills.HelpText = lipgloss.NewStyle().Foreground(fgSubtle) + s.Pills.QueueItemPrefix = lipgloss.NewStyle().Foreground(o.fgMoreSubtle).SetString(" •") + s.Pills.QueueItemText = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Pills.QueueLabel = lipgloss.NewStyle().Foreground(o.fgBase) + s.Pills.QueueIconBase = lipgloss.NewStyle().Foreground(o.fgBase) + s.Pills.QueueGradFromColor = o.error + s.Pills.QueueGradToColor = o.secondary + s.Pills.TodoLabel = lipgloss.NewStyle().Foreground(o.fgBase) + s.Pills.TodoProgress = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Pills.TodoCurrentTask = lipgloss.NewStyle().Foreground(o.fgMostSubtle) + s.Pills.TodoSpinner = lipgloss.NewStyle().Foreground(o.successMostSubtle) + s.Pills.HelpKey = lipgloss.NewStyle().Foreground(o.fgMoreSubtle) + s.Pills.HelpText = lipgloss.NewStyle().Foreground(o.fgMostSubtle) s.Pills.Area = base return s diff --git a/internal/ui/styles/styles.go b/internal/ui/styles/styles.go index ecde74258c127992e5a995adf74a69588cefe7a3..b8841e5c2881d25006840eb61d339bd104024901 100644 --- a/internal/ui/styles/styles.go +++ b/internal/ui/styles/styles.go @@ -17,10 +17,11 @@ import ( ) const ( - CheckIcon string = "✓" - SpinnerIcon string = "⋯" - LoadingIcon string = "⟳" - ModelIcon string = "◇" + CheckIcon string = "✓" + SpinnerIcon string = "⋯" + LoadingIcon string = "⟳" + ModelIcon string = "◇" + HypercreditIcon string = "◆" ArrowRightIcon string = "→" @@ -63,6 +64,7 @@ type Styles struct { Charm lipgloss.Style // Style for "Charm™" label Diagonals lipgloss.Style // Style for diagonal separators (╱) Percentage lipgloss.Style // Style for context percentage + Hypercredit lipgloss.Style // Style for Hypercredit count (◆ N) Keystroke lipgloss.Style // Style for keystroke hints (e.g., "ctrl+d") KeystrokeTip lipgloss.Style // Style for keystroke action text (e.g., "open", "close") WorkingDir lipgloss.Style // Style for current working directory @@ -189,6 +191,8 @@ type Styles struct { TokenCount lipgloss.Style // "(42K)" token count TokenPercentage lipgloss.Style // "42%" percent of context window Cost lipgloss.Style // "$0.42" cost readout + HypercreditIcon lipgloss.Style // Hypercredit icon (◆) + HypercreditText lipgloss.Style // Remaining Hypercredits text } // Resource styles the LSP/MCP/skills sidebar lists: their heading, diff --git a/internal/ui/styles/themes.go b/internal/ui/styles/themes.go index 39782c32840e44ebd4b134432be2d7df7fcd99a8..754abee7b4329f7237fced66340f4f129cbd59f3 100644 --- a/internal/ui/styles/themes.go +++ b/internal/ui/styles/themes.go @@ -2,40 +2,86 @@ package styles import "github.com/charmbracelet/x/exp/charmtone" +// ThemeForProvider returns the Styles associated with the given provider +// ID. Unknown or empty provider IDs yield the default Charmtone Pantera +// theme. +func ThemeForProvider(providerID string) Styles { + switch providerID { + case "hyper": + return HypercrushObsidiana() + default: + return CharmtonePantera() + } +} + // CharmtonePantera returns the Charmtone dark theme. It's the default style // for the UI. func CharmtonePantera() Styles { return quickStyle(quickStyleOpts{ primary: charmtone.Charple, secondary: charmtone.Dolly, - tertiary: charmtone.Bok, - - fgBase: charmtone.Ash, - fgMuted: charmtone.Squid, - fgHalfMuted: charmtone.Smoke, - fgSubtle: charmtone.Oyster, - - onPrimary: charmtone.Salt, - onAccent: charmtone.Butter, - - bgBase: charmtone.Pepper, - bgBaseLighter: charmtone.BBQ, - bgSubtle: charmtone.Charcoal, - bgOverlay: charmtone.Iron, - - border: charmtone.Charcoal, - borderFocus: charmtone.Charple, - - danger: charmtone.Coral, - error: charmtone.Sriracha, - warning: charmtone.Zest, - warningStrong: charmtone.Mustard, - busy: charmtone.Citron, - info: charmtone.Malibu, - infoSubtle: charmtone.Sardine, - infoMuted: charmtone.Damson, - success: charmtone.Julep, - successSubtle: charmtone.Bok, - successMuted: charmtone.Guac, + accent: charmtone.Bok, + keyword: charmtone.Blush, + + fgBase: charmtone.Ash, + fgMoreSubtle: charmtone.Squid, + fgSubtle: charmtone.Smoke, + fgMostSubtle: charmtone.Oyster, + + onPrimary: charmtone.Butter, + + bgBase: charmtone.Pepper, + bgLeastVisible: charmtone.BBQ, + bgLessVisible: charmtone.Charcoal, + bgMostVisible: charmtone.Iron, + + separator: charmtone.Charcoal, + + destructive: charmtone.Coral, + error: charmtone.Sriracha, + warningSubtle: charmtone.Zest, + warning: charmtone.Mustard, + busy: charmtone.Citron, + info: charmtone.Malibu, + infoMoreSubtle: charmtone.Sardine, + infoMostSubtle: charmtone.Damson, + success: charmtone.Julep, + successMoreSubtle: charmtone.Bok, + successMostSubtle: charmtone.Guac, + }) +} + +// HypercrushObsidiana returns the Hypercrush dark theme. +func HypercrushObsidiana() Styles { + return quickStyle(quickStyleOpts{ + primary: charmtone.Charple, + secondary: charmtone.Dolly, + accent: charmtone.Bok, + + fgBase: charmtone.Ash, + fgMoreSubtle: charmtone.Squid, + fgSubtle: charmtone.Smoke, + fgMostSubtle: charmtone.Oyster, + + onPrimary: charmtone.Butter, + + bgBase: charmtone.Pepper, + bgLeastVisible: charmtone.BBQ, + bgLessVisible: charmtone.Charcoal, + bgMostVisible: charmtone.Iron, + + separator: charmtone.Charcoal, + + destructive: charmtone.Coral, + error: charmtone.Sriracha, + warningSubtle: charmtone.Zest, + warning: charmtone.Mustard, + busy: charmtone.Citron, + info: charmtone.Malibu, + infoMoreSubtle: charmtone.Sardine, + infoMostSubtle: charmtone.Damson, + success: charmtone.Julep, + successMoreSubtle: charmtone.Bok, + successMostSubtle: charmtone.Guac, }) } diff --git a/schema.json b/schema.json index f60fd4f5f696f2d5f17aa4d5b2af4d6089336722..9d610579b90dedb636522532b3648d441fc0057f 100644 --- a/schema.json +++ b/schema.json @@ -102,10 +102,7 @@ } }, "additionalProperties": false, - "type": "object", - "required": [ - "tools" - ] + "type": "object" }, "HookConfig": { "properties": { @@ -354,8 +351,7 @@ "context_window", "default_max_tokens", "can_reason", - "supports_attachments", - "options" + "supports_attachments" ] }, "ModelOptions": { @@ -708,10 +704,7 @@ } }, "additionalProperties": false, - "type": "object", - "required": [ - "completions" - ] + "type": "object" }, "Token": { "properties": { @@ -779,11 +772,7 @@ } }, "additionalProperties": false, - "type": "object", - "required": [ - "ls", - "grep" - ] + "type": "object" } } }