code-review

Code Review Skill

Automatically reviews code quality when significant changes are made.

When to Activate

This skill should activate when:

• User writes or modifies 50+ lines of code
• User implements a new feature or handler
• User asks for code review or feedback
• Changes touch authentication, payments, or security

Review Dimensions

1. Security (Critical)

• No hardcoded secrets or credentials
• Input validation on all user inputs
• SQL injection / NoSQL injection prevention
• XSS prevention in any output
• Auth checks on protected endpoints

2. Project Patterns (Critical)

• Generated types used (not map[string]interface{})
• Routes registered properly
• Response schemas defined correctly
• OpenAPI spec updated for API changes

3. Performance

• No N+1 queries
• Appropriate indexes for queries
• Reasonable pagination limits
• No unnecessary data fetching

4. Architecture

• Single responsibility principle
• Proper error handling
• Appropriate logging
• Testable design

5. Platform Specific

Backend (Go)

• Uses Echo framework patterns
• Proper middleware usage
• Structured logging with Zap

iOS (Swift)

• SwiftUI best practices
• Source files < 400 lines
• Proper state management

Android (Kotlin)

• Jetpack Compose patterns
• Proper DI configuration
• Feature parity with iOS

Web (TypeScript)

• Next.js conventions followed
• Proper typing (no any)
• React best practices

Output Format

## Code Review Summary

### Critical Issues
- [List any blocking issues]

### Warnings
- [List non-blocking concerns]

### Suggestions
- [List optional improvements]

### Pattern Compliance
- [Check against CLAUDE.md rules]

Reference

See .claude/commands/code-review.md for detailed review criteria.