review-pr-comments
Skill0 starsChanged 27 days agoMITView on GitHub
Triage PR review comments in plan mode — investigate each one, mark invalid ones with a reason, plan fixes for valid ones, and compile a rejection list for the reviewer.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- IBazylchuk/paparats-mcptheir own version, first seen Nov 1, 2026
- fritzprix/libr-agenttheir own version, first seen Oct 26, 2026
- learntocloud/learn-to-cloud-apptheir own version, first seen Oct 8, 2026
- latitude-dev/latitude-llmtheir own version, first seen Oct 8, 2026
- DataDog/datadog-agenttheir own version, first seen Oct 8, 2026
- Hey-Diga/dotclaudetheir own version, first seen Feb 23, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- latitude-dev/latitude-llm★ 4.7ksame name
- DataDog/datadog-agent★ 3.8ksame name
- learntocloud/learn-to-cloud-app★ 1.1ksame name
- IBazylchuk/paparats-mcp★ 34same name
- Hey-Diga/dotclaude★ 17same name
- fritzprix/libr-agent★ 11same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository13 other files it gives its agents.
- arch-reviewSkill
- explain-diffSkill
- explainSkill
- find-sessionSkill
- prep-compactSkill
- prep-prSkill
- prune-commentsSkill
- remove-commentsSkill
- resume-planSkill
- review-pr-comments-loopSkill
- runbookSkill
- session-review-loopSkill
- session-reviewSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 3 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 23
Call EnterPlanMode immediately.
- Distinctive: line 64
If every comment is valid, write "All comments are valid — no rejections." in Section B.
- Very specific: line 72
Call ExitPlanMode with a plan built from Section A's Valid comments only, using their fix steps verbatim.
- Very specific: line 74
If every comment was Invalid, skip ExitPlanMode and report that no fixes are needed — do not exit plan mode with an empty plan.
- Very specific: line 78
Do not make any code edits before the plan is approved via ExitPlanMode — investigation and planning only.
- Distinctive: line 79
Do not invent fixes for comments you have not investigated.
- Distinctive: line 80
Keep rebuttals factual and collegial.
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Purpose
Triage a list of PR review comments. For each comment, investigate the actual codebase state to determine whether the comment is valid. Produce a plan that addresses valid comments and clearly rejects invalid ones, then summarise all rejections for the reviewer.
Input
The PR review comments are either:
- Provided as arguments after the slash command (e.g.
/review-pr-comments <pasted text>), or - The most recent block of numbered/bulleted text in the conversation.
Preserve the original numbering or bullet labels throughout.
Steps
1. Enter plan mode
Call EnterPlanMode immediately. All analysis and planning happens inside plan mode.
2. Parse the comments
Extract each distinct review comment from the input. Assign it a number that matches the original (1, 2, 3 … or the bullet label used by the reviewer). If the input has no explicit numbers, number them sequentially.
3. Investigate each comment
For every comment, read the relevant files and grep for the symbols, patterns, or logic the reviewer mentions. Do not rely on memory or assumptions — verify against the current codebase state.
Determine the verdict:
- Valid — the issue the reviewer describes genuinely exists in the current code. Plan a concrete fix.
- Invalid — the code already handles this correctly, the reviewer misread the code, the concern does not apply to this codebase, or the suggested change would be incorrect/harmful. Write a one- or two-sentence rebuttal that references the specific file and line (or pattern) that disproves the comment.
4. Structure the plan
Produce a plan document with two sections:
Section A — Comment-by-comment assessment
For each comment, in original order:
## Comment N — <one-line summary of the reviewer's point>
**Verdict:** Valid | Invalid
**Reason:** <for Invalid: rebuttal with file:line evidence>
**Fix steps:** <for Valid only: concrete, ordered steps — files to touch, functions/methods affected, the exact change to make. Write this so it can be executed directly, not summarized further.>Section B — Rejection list (for the reviewer)
A numbered list containing only the Invalid comments, using their original numbers, with a one- or two-sentence plain-language rebuttal for each. This section is ready to copy-paste back to the reviewer.
### Comments we are not actioning
N. <rebuttal>
N. <rebuttal>If every comment is valid, write "All comments are valid — no rejections." in Section B.
5. Save the reviewer rejection list
Save Section B alone to ~/.claude/plans/pr-review-triage-<YYYY-MM-DD>.md and state the filename. This file is for copy-pasting back to the reviewer — it is not the execution plan.
6. Exit plan mode with the fix plan
Call ExitPlanMode with a plan built from Section A's Valid comments only, using their fix steps verbatim. This is the plan the user approves to go straight into execution — it must contain only actionable fix steps, no rebuttals, no invalid-comment discussion.
If every comment was Invalid, skip ExitPlanMode and report that no fixes are needed — do not exit plan mode with an empty plan.
Constraints
- Do not make any code edits before the plan is approved via
ExitPlanMode— investigation and planning only. - Do not invent fixes for comments you have not investigated.
- Keep rebuttals factual and collegial; cite file paths and line numbers where they strengthen the case.
- Fix steps must be concrete enough to execute without re-deriving the approach — name the files, functions, and the change itself.
- Adhere to the user's
%%annotation convention: if%%markers appear in the plan file, address them before finalising.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

