agentleFS
rhvdberghrhvdbergh/claude-skills/review-pr-comments/SKILL.md

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

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
  1. Very specific:

    Call EnterPlanMode immediately.

    line 23
  2. Distinctive:

    If every comment is valid, write "All comments are valid — no rejections." in Section B.

    line 64
  3. Very specific:

    Call ExitPlanMode with a plan built from Section A's Valid comments only, using their fix steps verbatim.

    line 72
  4. Very specific:

    If every comment was Invalid, skip ExitPlanMode and report that no fixes are needed — do not exit plan mode with an empty plan.

    line 74
  5. Very specific:

    Do not make any code edits before the plan is approved via ExitPlanMode — investigation and planning only.

    line 78
  6. Distinctive:

    Do not invent fixes for comments you have not investigated.

    line 79
  7. Distinctive:

    Keep rebuttals factual and collegial.

    line 80

“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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

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.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.