From 565af03619ead5d0b0e17fd4922b6f3ad253db77 Mon Sep 17 00:00:00 2001 From: Amolith Date: Mon, 23 Mar 2026 09:23:51 -0600 Subject: [PATCH] refactor(comments): use CR instead of AGENT --- README.md | 27 +++++------ skills/addressing-agent-comments/SKILL.md | 47 ------------------- .../addressing-code-review-comments/SKILL.md | 45 ++++++++++++++++++ 3 files changed, 58 insertions(+), 61 deletions(-) delete mode 100644 skills/addressing-agent-comments/SKILL.md create mode 100644 skills/addressing-code-review-comments/SKILL.md diff --git a/README.md b/README.md index 0ca467e01c229946dbaff8c58842fdd4f9287586..0fe4531fa40353efdc4f20123a2509f92956ec7b 100644 --- a/README.md +++ b/README.md @@ -24,8 +24,7 @@ token count, plus overall metadata usage. I've used and tested them most with [Pi]: https://github.com/badlogic/pi-mono/tree/main/packages/coding-agent - [addressing-agent-comments](skills/addressing-agent-comments/SKILL.md): Finds - `AGENT:` comments in code, gathers feedback, and carries out requested - changes. + `CR:` comments in code, gathers feedback, and carries out requested changes. - [ast-grep](skills/ast-grep/SKILL.md): Writes [ast-grep] rules for structural code search and modification. - [auditing-repositories](skills/auditing-repositories/SKILL.md): Audits open @@ -235,14 +234,14 @@ endpoint with [./skill-stats.go](./skill-stats.go). [/messages/count_tokens]: https://dev.synthetic.new/docs/anthropic/messages/count-tokens ``` -=== addressing-agent-comments === +=== addressing-code-review-comments === Token breakdown: - Name: 8 tokens - Description: 36 tokens - Body: 271 tokens (40 lines) + Name: 9 tokens + Description: 40 tokens + Body: 290 tokens (38 lines) ─────────────────────────────────────────────── - Total: 315 tokens + Total: 339 tokens === ast-grep === @@ -439,9 +438,9 @@ Token breakdown: Token breakdown: Name: 10 tokens Description: 60 tokens - Body: 1152 tokens (111 lines) + Body: 1217 tokens (112 lines) ─────────────────────────────────────────────── - Total: 1222 tokens + Total: 1287 tokens === resuming-work-through-lunatask === @@ -508,9 +507,9 @@ Token breakdown: Token breakdown: Name: 8 tokens Description: 32 tokens - Body: 607 tokens (88 lines) + Body: 909 tokens (128 lines) ─────────────────────────────────────────────── - Total: 647 tokens + Total: 949 tokens === writing-git-tags === @@ -552,9 +551,9 @@ SUMMARY ============================================================ Skills: 29 -Metadata: 1765 tokens -Combined bodies: 29737 tokens -Overall: 83884 tokens +Metadata: 1770 tokens +Combined bodies: 30123 tokens +Overall: 84275 tokens Validation errors: 0 Largest skills (by total tokens): diff --git a/skills/addressing-agent-comments/SKILL.md b/skills/addressing-agent-comments/SKILL.md deleted file mode 100644 index 8d876aaa945b7002314776425aa6778dd06f150e..0000000000000000000000000000000000000000 --- a/skills/addressing-agent-comments/SKILL.md +++ /dev/null @@ -1,47 +0,0 @@ ---- -name: addressing-agent-comments -description: Finds AGENT comments in code, gathers feedback, and carries out requested changes. Use when scanning a repo for inline agent notes, decision requests, or tooling tags. -metadata: - source: https://github.com/ampcode/amp-contrib ---- - -Search for inline `AGENT:` comments, read the surrounding code, and follow through on the request. - -## When to use - -- You need to collect and resolve inline agent notes or change requests -- A repo uses a tag like AGENT: to point at work items -- You are asked to scan for inline feedback embedded in code - -## Search process - -1. Identify the instruction tag for the repo. Default to `AGENT:` if none is provided. -2. Use ripgrep to find the tag with context: - -```bash -rg -n --fixed-strings "AGENT:" -C 3 -``` - -3. If multiple tags are used, repeat the search for each tag. - -## Review process - -- Read each comment with surrounding code to understand scope and intent -- Make the change or address the concern where feasible -- If a request is unclear, document assumptions and raise questions -- Update or remove resolved AGENT: comments - -## Output format - -- Group findings by file path -- Include line numbers and relevant context for each comment -- Summarize key themes and action items at the end - -## Expected actions - -After finding AGENT: comments: - -1. Analyze the request in each comment -2. Make the change, or explain why you did not -3. Resolve or update the instruction comment -4. Provide a brief summary of actions taken and any open questions diff --git a/skills/addressing-code-review-comments/SKILL.md b/skills/addressing-code-review-comments/SKILL.md new file mode 100644 index 0000000000000000000000000000000000000000..3ce5710e429e92abe75762500469346e17ba9b41 --- /dev/null +++ b/skills/addressing-code-review-comments/SKILL.md @@ -0,0 +1,45 @@ +--- +name: addressing-code-review-comments +description: Finds 'CR:' comments in code, gathers feedback, and carries out requested changes. Use when scanning a repo for inline agent notes, code review from the user, or decision requests. +metadata: + source: https://github.com/ampcode/amp-contrib +--- + +Search for inline `CR:` comments, read the surrounding code, and follow through on the request. + +## When to use + +- You need to collect and resolve inline agent notes or change requests +- A repo uses a tag like CR: to point at work items +- You are asked to scan for inline feedback embedded in code + +## Search process + +1. Identify the instruction tag for the repo. Default to `CR:` if none is provided. +2. Use ripgrep to find the tag with context: + ```bash + rg -n --fixed-strings "CR:" -C 3 + ``` +3. If multiple tags are used, repeat the search for each tag. + +## Review process + +- Read each comment with surrounding code to understand scope and intent, maybe the whole file is relevant +- Make the change, address the concern, or answer the question where feasible +- If a request is unclear, stop and ask the user to elaborate +- Update or remove resolved CR: comments + +## Output format + +- Group findings by file path +- Include line numbers and relevant context for each comment +- Summarize key themes and action items at the end + +## Expected actions + +After finding CR: comments: + +1. Analyze the request in each comment +2. Make the change or explain why you did not, or answer the question, or otherwise address the comment +3. Resolve or update the comment +4. Provide a brief summary of actions taken and any open questions