Use `HYPER_API_KEY` to configure Hyper for non-interactive workflows and
bypass OAuth-driven setup when the environment variable is present.

💘 Generated with Crush

Assisted-by: Kimi K2.5 via Crush <crush@charm.land>

Ensures crush_info reflects fresh config after edits and detects
external/manual config file modifications.

Track config file staleness and expose dirty/changed/missing state so
crush_info can warn when in-memory config is outdated after on-disk
edits. This also makes reload use freshly swapped config state before
model/agent setup to avoid stale runtime wiring.

💘 Generated with Crush

Assisted-by: Kimi-K2.5 via Crush <crush@charm.land>

Implements a tool that reads Crush's own log files and returns
formatted entries in compact text format. Supports:

- Tail-style reading (last N entries, default 50, max 100)
- Sensitive value redaction (token, secret, api_key, etc.)
- JSON value formatting with proper escaping
- Chronological ordering (oldest first)
- Graceful handling of missing/empty files and malformed lines

Approximate overhead:

• Tool definition cost (always in prompt): ~1,000 bytes ≈ 250 tokens
• Per invocation response (50 lines): ~4,000 bytes ≈ 1,000 tokens
• Per invocation response (100 lines): ~8,000 bytes ≈ 2,000 tokens

💖 Generated with Crush

Assisted-by: Kimi-K2.5 via Crush <crush@charm.land>

Approximate overhead:

• Tool definition cost (always in prompt): ~500 bytes ≈ 130 tokens
• Per invocation response (typical): ~1,500 bytes ≈ 375 tokens
• Per invocation response (worst-case): ~3,000 bytes ≈ 750 tokens

💘 Generated with Crush

Assisted-by: Kimi-K2.5 via Crush <crush@charm.land>

Bumps the all group with 3 updates: [charm.land/catwalk](https://github.com/charmbracelet/catwalk), [github.com/go-git/go-git/v5](https://github.com/go-git/go-git) and [modernc.org/sqlite](https://gitlab.com/cznic/sqlite).


Updates `charm.land/catwalk` from 0.33.2 to 0.34.3
- [Release notes](https://github.com/charmbracelet/catwalk/releases)
- [Commits](https://github.com/charmbracelet/catwalk/compare/v0.33.2...v0.34.3)

Updates `github.com/go-git/go-git/v5` from 5.17.1 to 5.17.2
- [Release notes](https://github.com/go-git/go-git/releases)
- [Commits](https://github.com/go-git/go-git/compare/v5.17.1...v5.17.2)

Updates `modernc.org/sqlite` from 1.48.0 to 1.48.1
- [Changelog](https://gitlab.com/cznic/sqlite/blob/master/CHANGELOG.md)
- [Commits](https://gitlab.com/cznic/sqlite/compare/v1.48.0...v1.48.1)

---
updated-dependencies:
- dependency-name: charm.land/catwalk
  dependency-version: 0.34.3
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: all
- dependency-name: github.com/go-git/go-git/v5
  dependency-version: 5.17.2
  dependency-type: direct:production
  update-type: version-update:semver-patch
  dependency-group: all
- dependency-name: modernc.org/sqlite
  dependency-version: 1.48.1
  dependency-type: direct:production
  update-type: version-update:semver-patch
  dependency-group: all
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Fix LSP UI display broken in v0.55.0. The UI was still subscribing
to workspace.LSPEvent after the event type was moved to the app
package, causing LSP status to never update in the UI.

Changes:
- internal/ui/model/ui.go: Subscribe to app.LSPEvent; use app.GetLSPStates()
- internal/ui/model/lsp.go: Use app.LSPClientInfo type for LSPInfo embedding
  and sorting (fields are identical to workspace.LSPClientInfo)

Fixes charmbracelet/crush#2560

When Hyper returns a 401, publish a TypeReAuthenticate notification so the
UI opens the OAuth dialog. Also fix coordinator.isUnauthorized to recognize
hyper.ErrUnauthorized for auto-refresh.

💘 Generated with Crush

Assisted-by: Z.ai: GLM 5.1 via Crush <crush@charm.land>

Two fixes for local/custom model compatibility (LM Studio, Ollama, llama.cpp):

1. Don't send MaxOutputTokens when it's 0. Custom models not in the
   catwalk providers list have DefaultMaxTokens=0, which gets sent as
   max_tokens:0 in the API request. LM Studio rejects this with
   "maxPredictedTokens does not satisfy the schema". Fix: only send
   the field when the value is positive.

2. Skip auto-summarize when ContextWindow is 0. Custom models have
   ContextWindow=0, making remaining tokens negative, which immediately
   triggers summarize after the first response. The session resets with
   "previous session was interrupted because it got too long" even for
   short conversations. Fix: skip the check when context window is
   unknown.

Fixes #1218 (regression), relates to #1583, #1591

Feat/Refactor: Server Client architecture and API

See: https://git-scm.com/docs/gitignore

Co-authored-by: Andrey Nering <andreynering@users.noreply.github.com>

I found Crush was often both too verbose and too terse than I'd like in
different ways when generating the AGENTS.md. This is an attempt to
improve the relevancy of the information it mentions.

As just one example, tabs vs spaces is obvious from the first file read,
so mentioning that style preference in AGENTS.md just wastes tokens.
Since adding this line, I haven't noticed it including trivial details
like that!

Though I've been using this since Dec 20 and liking it, I've mostly used
it and had good results with Claude models. I've not tried it with GPT
or GLM or others yet.

The GetMCPPrompt function was using context.Background() without any
timeout, which could cause the request to hang indefinitely. Since
tea.Cmd doesn't support context passing, we add a 30-second timeout
as a safeguard, consistent with the pattern used in OAuth modules.

Assisted-by: GLM-5 via Crush <crush@charm.land>