skills/gannonh/skills/coderabbit-review

coderabbit-review

SKILL.md

CodeRabbit Review Handler

Process CodeRabbit review comments with context-aware discretion.

Instructions

1. Initial Context

Inform the user:

I'll review the CodeRabbit comments with discretion, as CodeRabbit doesn't have access to the entire codebase and may not understand the full context.

For each comment, I'll:
- Evaluate if it's valid given our codebase context
- Accept suggestions that improve code quality
- Ignore suggestions that don't apply to our architecture
- Explain my reasoning for accept/ignore decisions

2. Run Code Rabbit

If no comments are provided, run CodeRabbit in prompt-only mode to generate comments:

# Run CodeRabbit cli
coderabbit --prompt-only 

3. Evaluate Comments

  • Parse the comments
  • For each comment:
    • Read the relevant file to understand context
    • Determine if the suggestion is valid and beneficial
    • Decide to accept or ignore the suggestion
    • Document reasoning for each decision
  • Ignore *.md files

4. Execute Fixes

Task a subagent to address the comments you've accepted

5. Validate Changes

Run relevant tests and linters to ensure code integrity:

Example:

# Run all unit tests
./scripts/test.sh unit 1 2>&1 | tail -100

# Run affected UI tests
./scripts/test.sh ui [TestName] 2>&1 | tail -100

# Run swiftlint on changed files and fix ANY issues
git diff --name-only origin/main...HEAD -- '*.swift' | xargs -r swiftlint lint --strict

6. Commit Changes

After applying changes commit your work using appropriate commit messages summarizing the changes made. Address any re-commit hook violations as needed.

7. Consolidate Results

After completion, provide a summary report:

📋 CodeRabbit Review Summary

Files Processed: {count}

Accepted Suggestions:
  {file}: {changes_made}
  
Ignored Suggestions:
  {file}: {reason_ignored}

Overall: {X}/{Y} suggestions applied

Common Patterns to Ignore

  • Style preferences that conflict with project conventions
  • Generic best practices that don't apply to our specific use case
  • Performance optimizations for code that isn't performance-critical
  • Accessibility suggestions for internal tools
  • Security warnings for already-validated patterns
  • Import reorganization that would break our structure

Common Patterns to Accept

  • Actual bugs (null checks, error handling)
  • Security vulnerabilities (unless false positive)
  • Resource leaks (unclosed connections, memory leaks)
  • Type safety issues (TypeScript/type hints)
  • Logic errors (off-by-one, incorrect conditions)
  • Missing error handling

Decision Framework

For each suggestion, consider:

  1. Is it correct? - Does the issue actually exist?
  2. Is it relevant? - Does it apply to our use case?
  3. Is it beneficial? - Will fixing it improve the code?
  4. Is it safe? - Could the change introduce problems?

Only apply if all answers are "yes" or the benefit clearly outweighs risks.

Important Notes

  • CodeRabbit is helpful but lacks context
  • Trust your understanding of the codebase over generic suggestions
  • Explain decisions briefly to maintain audit trail
  • Batch related changes for efficiency
  • Always run the relevant ui and unit tests, and/or create new tests to verify correctness after applying changes
Weekly Installs
24
Repository
gannonh/skills
First Seen
Feb 18, 2026
Installed on
opencode24
claude-code24
github-copilot24
codex24
kimi-cli24
gemini-cli24