OpenParable
Skip to content
← Back to skills

Review Comments

Testing & Quality
review_comments

Review Comments is an Agent Skill for testing & quality published in the pgplex/pgschema repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Verdict PR review comments pragmatically - address valid ones, dismiss over-engineered or incorrect ones, and respond to each. Use this skill when handling PR review feedback, code review comments, or when asked to address reviewer suggestions.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/pgplex/pgschema/blob/HEAD/.claude/skills/review_comments/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/review-comments-90f931be, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Review Comments

Pragmatically triage PR review comments: fix what matters, push back on what doesn't.

Phase 1: Fetch Comments

  1. Get the PR number from context or user input
  2. Fetch all review comments:
    gh pr view <number> --json reviews,comments
    gh api repos/{owner}/{repo}/pulls/<number>/comments
    gh api repos/{owner}/{repo}/pulls/<number>/reviews
    
  3. Read the actual code being reviewed to understand context

Phase 2: Verdict Each Comment

For each comment, classify it into one of:

ADDRESS - The comment is correct and actionable

  • Genuine bugs or logic errors
  • Missing edge cases that could actually happen
  • Security issues
  • Correctness issues backed by evidence

DISMISS - The comment is over-engineered, stylistic, or wrong

  • Adding abstractions for hypothetical future needs
  • Suggesting error handling for impossible scenarios
  • Style preferences with no functional impact
  • Refactoring suggestions unrelated to the PR's purpose
  • Incorrect assumptions about how the code works
  • "What if..." scenarios that won't happen given the codebase constraints
  • Suggesting tests for trivial code paths
  • Adding comments/docs to self-evident code

ACKNOWLEDGE - Minor valid point, not worth changing

  • Technically correct but the cost of change outweighs the benefit
  • Valid style point but not a blocker
  • Good idea for a follow-up, not this PR

Phase 3: Act on Each Comment

For ADDRESS comments

  1. Make the code change
  2. Reply explaining what was fixed

For DISMISS comments

Reply with a concise, respectful explanation of why you're not making the change. Be direct, not defensive. Examples:

  • Over-engineering: "This is a single-use path - I'd rather keep it simple and extract if/when we need reuse."
  • Impossible scenario: "This can't happen because [X guarantees Y]. Adding a guard here would suggest the invariant isn't trustworthy."
  • Style preference: "Both approaches work - I'll keep the current style for consistency with the surrounding code."
  • Wrong assumption: "Actually, [correct explanation]. The current code handles this because [reason]."

For ACKNOWLEDGE comments

Reply acknowledging the point briefly: "Good point - I'll keep this in mind for follow-up" or "Fair, though I think the current approach is fine for now given [reason]."

Phase 4: Respond

Use gh to reply to each comment:

gh api repos/{owner}/{repo}/pulls/<number>/comments/<comment_id>/replies \
  -f body="<response>"

Decision Principles

  1. Ship over perfection - The goal is a working, correct change, not a perfect one
  2. YAGNI - Don't build for requirements that don't exist yet
  3. Minimal diff - Every line changed is a line that could break; keep changes focused
  4. Trust the codebase - Internal code doesn't need defensive programming against itself
  5. Reviewer != oracle - Reviewers can be wrong, overly cautious, or applying rules from different contexts
  6. Evidence over authority - "This could break" needs a scenario; "best practice says" needs a reason
  7. Be respectful, be direct - Dismissing a comment isn't personal; explain your reasoning clearly