CodeRabbit Review

Run CodeRabbit CLI for external AI-powered code review.

Core principle: External validation complements internal reviews. CodeRabbit provides language-specific best practices; Ring reviewers understand your requirements.

When to Use

Use CodeRabbit review when:

• Want external perspective (different from Ring's internal reviewers)
• Need language-specific linting and best practices
• Want to validate against broader code pattern datasets
• Need additional security vulnerability scanning

Use alongside Ring's internal reviews:

• Internal reviews (code/business/security) - Ring agents in parallel, context-aware
• External review (CodeRabbit) - Independent validation, pattern-based

Don't use as replacement for:

• Ring's internal reviewers (use both, not either/or)
• Requirements validation (CodeRabbit doesn't know your PRD)
• Business logic verification (CodeRabbit doesn't understand domain)

Review Types

All changes (default):

/review-cr

Reviews both committed and uncommitted changes.

Committed only:

/review-cr committed

Reviews only committed changes (staged history).

Uncommitted only:

/review-cr uncommitted

Reviews only working directory changes (unstaged).

Custom base branch:

/review-cr --base develop
/review-cr committed --base main

Compare against specific branch instead of repository default.

The Process

1. Argument Parsing

Parse arguments in any order:

• Scan for --base flag: extract next argument as base branch
• Check for review type: uncommitted, committed, or all
• Defaults: type=all, base=repository default branch

2. Validation

Validate base branch if specified:

git rev-parse --verify <branch>

If branch doesn't exist:

• Message: "Base branch '' not found. Available branches: "
• Exit gracefully without running review

After validation, sanitize branch name:

branch_safe=$(echo "$branch" | sed 's/[^a-zA-Z0-9._/-]//g')

This prevents command injection via branch names.

Check for existing CodeRabbit processes:

pgrep -f "coderabbit review"

If process exists:

• Message: "A CodeRabbit review is already running. Wait for completion or kill process."
• Exit gracefully

3. Execute Review

Run in background:

coderabbit review --prompt-only --type <type> [--base <branch_safe>]

Use Bash tool with run_in_background: true.

CodeRabbit context sources:

• ~/.claude/CLAUDE.md - Your global coding standards
• <project>/.claude/claude.md - Project-specific guidelines
• <project>/.coderabbit.yaml - CodeRabbit configuration

These provide coding standards and preferences to the review.

4. Monitor Progress

Check every 30 seconds:

• Use BashOutput tool to monitor background process
• Look for final summary in output
• Maximum wait: 45 minutes (90 checks)

Progress messages:

• At 5 minutes: "CodeRabbit is still analyzing... (can take 7-30 minutes for large changesets)"
• Continue monitoring

If BashOutput fails:

• Wait 10 seconds and retry once
• If retry fails: "Unable to monitor review process. Check with: ps aux | grep coderabbit"

If timeout (45 minutes):

• Report: "Review exceeded 45 minute timeout. Terminating runaway process..."
• Graceful termination:

  pid=$(pgrep -f "coderabbit review")
  kill -TERM $pid
  sleep 5
  pgrep -f "coderabbit review" && kill -KILL $pid
  

• Verify termination: pgrep -f "coderabbit review" returns nothing
• Message: "CodeRabbit process terminated after timeout. Prevents resource exhaustion."

5. Parse and Present Findings

Extract issues from output:

• Group by severity: CRITICAL, HIGH, MEDIUM, LOW, INFO
• For each issue:
  • Severity badge (🔴 CRITICAL, 🟠 HIGH, 🟡 MEDIUM, 🔵 LOW, ⚪ INFO)
  • File path and line number
  • Issue description
  • Suggested fix approach (if provided)

Example output:

🔴 CRITICAL - src/auth.ts:42
JWT secret hardcoded in source code
Fix: Move to environment variable

🟠 HIGH - src/api/users.ts:156
SQL injection vulnerability in query builder
Fix: Use parameterized queries

🟡 MEDIUM - src/utils/validator.ts:89
Missing input validation for email field
Fix: Add email format validation

6. Create Todos for Critical Issues

Use TodoWrite for CRITICAL and HIGH severity:

Format: "Fix [severity] in [file]:[line] - [brief description]"

Example:

• "Fix CRITICAL in src/auth.ts:42 - Remove hardcoded JWT secret"
• "Fix HIGH in src/api/users.ts:156 - SQL injection vulnerability"

7. Prompt for Action

CodeRabbit review complete. Found X issues (Y critical, Z high).
Would you like me to fix these issues?

Error Handling

CodeRabbit CLI not installed:

Message: "CodeRabbit CLI not installed. Install with:
         curl -fsSL https://cli.coderabbit.ai/install.sh | sh"
Exit gracefully

Not authenticated:

Message: "Not authenticated with CodeRabbit. Run: coderabbit auth login"
Exit gracefully

No changes to review:

Message: "No changes found to review. Make some changes first."
Exit gracefully

Review takes longer than expected:

Message at 5 minutes: "CodeRabbit is still analyzing...
                       (this can take 7-30 minutes for large changesets)"
Continue monitoring

Integration with Ring Reviews

Recommended workflow:

1. Internal review first (fast, context-aware):

   Dispatch Ring's 3 reviewers in parallel:
   - code-reviewer
   - business-logic-reviewer
   - security-reviewer
   

2. External review second (validation, patterns):

   /review-cr
   

3. Compare findings:

   • Issues found by both → High confidence, fix immediately
   • Issues found only by Ring → Context-specific, trust internal
   • Issues found only by CodeRabbit → Pattern-based, evaluate relevance

4. Handle by severity:

   • Critical/High (any source) → Fix immediately
   • Medium → Fix if time permits
   • Low → Add TODO comments
   • Info → Consider for future improvements

Red Flags

Never:

• Use CodeRabbit as sole review method (complement Ring's reviewers)
• Skip validation of base branch (security risk)
• Ignore timeouts (can cause resource exhaustion)
• Run multiple CodeRabbit reviews simultaneously (process conflict)

Always:

• Sanitize branch names before using in commands
• Monitor for completion (don't fire and forget)
• Create todos for Critical/High issues
• Terminate runaway processes after timeout

Anti-Patterns

Wrong:

• "CodeRabbit found nothing, so we're good" → Still run Ring reviewers
• "Let's run both reviews in parallel" → Run Ring first (faster), then CodeRabbit
• "Skip CodeRabbit, Ring reviews are enough" → External validation catches different issues
• "CodeRabbit says fix, so fix blindly" → Evaluate if pattern applies to your context

Right:

• Use Ring reviewers for requirements/business logic
• Use CodeRabbit for language-specific patterns
• Combine findings from both sources
• Prioritize Critical/High from either source

Required Patterns

This skill uses these universal patterns:

• State Tracking: See skills/shared-patterns/state-tracking.md
• Failure Recovery: See skills/shared-patterns/failure-recovery.md
• Exit Criteria: See skills/shared-patterns/exit-criteria.md
• TodoWrite: See skills/shared-patterns/todowrite-integration.md

Apply ALL patterns when using this skill.