code-review

When asked for an intense code review, specifically an agent code review, you will delegate tasks to background agents called 'codex' and 'gemini' to review the code comperehensively and get back reports for action items and alternative perspectives.

⭐2stars🔀0forks📁View on GitHub🕐Updated Jan 10, 2026

When & Why to Use This Skill

This Claude skill automates comprehensive code reviews by orchestrating parallel sub-agents to analyze code for bugs, spec compliance, and architectural improvements. It solves the problem of manual review fatigue and oversight by providing multi-perspective feedback and actionable reports, ensuring higher code quality and faster merge cycles.

Use Cases

  • Pre-merge Quality Gate: Automatically perform a deep, multi-perspective review of pull requests to identify critical bugs and regressions before they reach production.
  • Compliance and Standards Enforcement: Ensure all new code adheres to internal organizational standards and specific project specifications through specialized sub-agent checks.
  • API and Library Validation: Use background agents to verify that external library calls and API integrations are implemented correctly according to the latest documentation.
  • Architectural Optimization: Get alternative implementation perspectives and recommendations for more efficient or maintainable code structures from specialized sub-agents.
namecode-review
descriptionComprehensive multi-agent code review with parallel analysis. Use for thorough code reviews, finding bugs, checking regressions, or getting alternative implementation perspectives.
contextfork

Agent Code Review

Orchestrate parallel sub-agent reviews for comprehensive code analysis and actionable feedback.

When to Use

  • Deep code review before merge
  • Bug hunting and regression detection
  • Getting alternative implementation perspectives
  • Validating library/API usage correctness

Instructions

Step 1: Determine Review Scope

Identify what to review:

# Option A: Recent changes
git diff HEAD~1

# Option B: Staged changes  
git diff --staged

# Option C: Specific commit range
git log --oneline -10  # then select range

# Option D: Current working state
git status

Step 2: Spawn Review Sub-Agents

Deploy parallel sub-agents with focused tasks:

Sub-Agent Task
Spec Reviewer Check implementation against UPP-1.3.spec.md and compliance requirements
Standards Reviewer Compare against existing provider patterns and org coding standards
Bug Hunter Search for regressions, edge cases, error handling gaps
API Validator Web search to verify API/library usage correctness

Prompt each sub-agent with:

"Perform a comprehensive code review of [scope]. Investigate spec compliance, coding standards consistency, and validate any API/library usage against current documentation. Report findings as actionable items."

Step 3: Collect and Synthesize

Aggregate sub-agent reports into:

  1. Critical issues requiring immediate fix
  2. Recommendations for improvement
  3. Alternative approaches worth considering
  4. Regression risks with resolution plans

Output Format

# Code Review Report

## Files Reviewed
- `path/to/file1.ts`
- `path/to/file2.ts`

## Critical Issues 🔴
| File | Line | Issue | Fix |
|------|------|-------|-----|

## Recommendations 🟡
| File | Line | Suggestion | Rationale |
|------|------|------------|-----------|

## Alternative Perspectives 🔵
[Alternative implementation approaches from sub-agents]

## Regression Analysis
| Risk | Affected Area | Resolution Plan |
|------|---------------|-----------------|

## Action Items
- [ ] Fix: [critical item]
- [ ] Consider: [recommendation]
- [ ] Verify: [needs validation]

Notes

  • Sub-agents should search the codebase for similar patterns before suggesting changes
  • Always create resolution plans for identified regressions
  • Validate external dependencies against their official documentation