Address PR Review Feedback

VerifiedSafe

Reads all review comments on a pull request, independently verifies each suggestion's correctness, applies or rejects them with justification, and posts a summary before re-requesting review. Best used when you need to critically evaluate reviewer feedback and ensure changes are safe and correct.

Sby Skills Guide Bot
DevelopmentIntermediate
6/2/2026
Claude CodeCursorWindsurf
#pr-review#feedback#code-review#verification#collaboration

Recommended for

Automated summary

This skill helps an AI agent address pull request review feedback by independently verifying each comment, applying or rejecting changes based on correctness, and re-requesting review.

Strengths

  • Independently verifies feedback before applying, reducing blind acceptance of incorrect suggestions.
  • Categorizes feedback by severity and tracks every comment.
  • Provides a structured response and summary for re-review.

Limitations

  • Requires the agent to have access to the repository and ability to run a local dev server.
  • May not cover all edge cases of visual testing or security logic.
  • Depends on the agent's ability to reason about code correctness.
When to use it

Use when you need to systematically respond to a reviewed pull request and ensure only valid feedback is applied.

When not to use it

Avoid for trivial PRs with no feedback or when a human must make subjective design decisions.

Security analysis

Safe
Quality score92/100

The skill uses common development commands (npm, npx, gh) for legitimate PR review purposes. It emphasizes independent verification of feedback before applying changes, and does not instruct any destructive or exfiltrating actions. While powerful tools are used, the risk is minimal in a controlled development environment.

No concerns found

How this analysis is made

Examples

Address review comments on a PR
Address review comments on PR #15 with independent verification.
Process and respond to feedback
Process and respond to feedback on pull request #42 using the address-review skill.
Verify and apply review suggestions
Verify and apply review suggestions for PR #123, handling escalations.

name: address-review description: Address PR review feedback, verify independently, and re-request review disable-model-invocation: true argument-hint: <pr-number>

Address Review Feedback

Address review comments on PR #$ARGUMENTS. Be adversarial about the feedback itself — verify that suggestions are correct before applying them.

PR Context

!gh pr view $ARGUMENTS
!gh api repos/{owner}/{repo}/pulls/$ARGUMENTS/comments --jq '.[] | "---\n\(.path):\(.line // .original_line)\n\(.body)\n"'
!gh pr view $ARGUMENTS --comments

Instructions

Step 1: Gather All Feedback

Read every review comment above. Categorize each piece of feedback:

  • Action Required — Reviewer flagged as blocking merge
  • Recommended — Reviewer suggested addressing before merge
  • Minor — Nits, style suggestions

Step 2: Independently Verify Each Finding

Do not blindly apply suggestions. For each piece of feedback:

  1. Read the file in question — Use Glob/Grep/Read to find the relevant file and understand the full context, not just the diff snippet.
  2. Assess whether the feedback is correct — Reviewers (including agent reviewers) can be wrong. Check:
    • Does the suggested change actually fix the issue identified?
    • Could the suggestion introduce a new bug or visual regression?
    • Is the reviewer missing context that makes the current code correct?
    • Does the suggestion align with project conventions in AGENTS.md?
  3. For visual feedback — Check out the branch, run npm run dev, verify rendering at mobile and desktop widths, check both variants (/start and /build).
  4. For security feedback — Take it seriously by default. Security suggestions should be applied unless you can clearly demonstrate they're wrong.

Step 3: Respond to Each Finding

For each piece of feedback, take one of these actions:

Apply — The feedback is correct. Make the change.

  • Edit the file
  • Run npm run dev and verify both variants render correctly
  • Note what was changed

Partially apply — The core insight is right but the suggested fix isn't quite right.

  • Implement a better fix that addresses the underlying concern
  • Explain why you deviated from the exact suggestion

Reject with justification — The feedback is incorrect or doesn't apply.

  • Explain clearly why the current code is correct
  • Reference AGENTS.md or project conventions to support your reasoning
  • Never reject feedback without a concrete justification

Escalate — You're unsure whether the feedback is valid.

  • Flag it to the human with the evidence for and against
  • Do not guess or silently skip

Step 4: Run Full Verification

After addressing all feedback:

  1. Run npm run dev and verify both variants render correctly at mobile and desktop widths
  2. Check TypeScript compilation — npx wrangler types && npx tsc --noEmit (Worker-side only)
  3. Test form submission if any API or form changes were made
  4. Spot-check your changes — Read through your own diff. Did you introduce any new issues while fixing the review feedback?
  5. Check sizing — If the fixes significantly expanded the PR, flag whether it should be split.

Step 5: Commit and Push

  • Commit fixes with clear messages linking to the review feedback: fix: Address review — <description>
  • Keep fix commits separate from each other when they address unrelated feedback (easier to review the re-review)
  • Push to the PR branch

Step 6: Post Summary and Re-request Review

Post a comment on the PR summarizing how each finding was addressed, then re-request review:

gh pr comment $ARGUMENTS --body "<summary>"
gh pr edit $ARGUMENTS --add-reviewer <reviewer>

Use this format for the summary:

## Review Feedback Addressed

| # | Finding | Action | Details |
|---|---------|--------|---------|
| 1 | <brief description> | Applied / Partially applied / Rejected | <what was done and why> |
| 2 | ... | ... | ... |

**Visual verification:** Both variants checked at mobile + desktop
**New commits:** <list of fix commits>

For any rejected findings, provide the full justification in the Details column so the reviewer can evaluate your reasoning.

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/address-pr-review-feedback-mm5t8c4y' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/address-pr-review-feedback-mm5t8c4y/download' --output '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md.part' && mv '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md.part' '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md' && test -s '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/address-review' && cp -R '.skill-downloads/address-pr-review-feedback-mm5t8c4y' '.claude/skills/address-review'

Cursor
test -f '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md' && test -s '.skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md' && mkdir -p '.cursor/skills' && test ! -e '.cursor/skills/address-review' && cp -R '.skill-downloads/address-pr-review-feedback-mm5t8c4y' '.cursor/skills/address-review'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/address-pr-review-feedback-mm5t8c4y'

openskills read 'address-pr-review-feedback-mm5t8c4y'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/address-pr-review-feedback-mm5t8c4y/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills