Agent Skills: Code Review Skill

Code Review Skill

A systematic approach to reviewing code for quality, security, and maintainability.

When to Use This Skill

• User asks to "review code" or "check my code"
• Reviewing pull requests or diffs
• Code audit tasks
• Pre-deployment code checks

Code Review Checklist

1. Functionality

• [ ] Does the code do what it's supposed to do?
• [ ] Are edge cases handled?
• [ ] Are error conditions handled appropriately?

2. Code Quality

• [ ] Is the code readable and self-documenting?
• [ ] Are variable/function names descriptive?
• [ ] Is there unnecessary code duplication?
• [ ] Is the code well-organized?

3. Security

• [ ] Input validation present?
• [ ] SQL injection prevention?
• [ ] XSS prevention for web code?
• [ ] Sensitive data properly handled?
• [ ] Authentication/authorization correct?

4. Performance

• [ ] No obvious performance issues?
• [ ] Efficient algorithms used?
• [ ] Database queries optimized?
• [ ] Appropriate caching?

5. Testing

• [ ] Unit tests included?
• [ ] Edge cases tested?
• [ ] Tests pass?

Review Approach

# Example: Reviewing a function
def review_function(code: str) -> dict:
    """Analyze a function for common issues."""
    issues = []
    suggestions = []
    
    # Check function length
    lines = code.strip().split('\n')
    if len(lines) > 50:
        issues.append("Function is too long (>50 lines). Consider breaking it up.")
    
    # Check for docstring
    if '"""' not in code and "'''" not in code:
        issues.append("Missing docstring. Add documentation.")
    
    # Check for bare except
    if 'except:' in code:
        issues.append("Bare except clause. Specify exception types.")
    
    # Check for hardcoded values
    import re
    if re.search(r'["\'](?:password|secret|key)["\']', code, re.I):
        issues.append("Possible hardcoded credential. Use environment variables.")
    
    return {"issues": issues, "suggestions": suggestions}

Feedback Format

When providing review feedback:

1. Start positive - Note what's done well
2. Be specific - Point to exact lines/issues
3. Explain why - Not just what's wrong, but why it matters
4. Suggest fixes - Provide concrete improvements

Example Feedback Template:

## Code Review Summary

### ✅ What's Good
- Clear function naming
- Good error handling in X

### 🔧 Suggested Improvements
1. **Line 25**: Consider using a constant for the magic number `42`
   - Why: Improves readability and maintainability
   - Suggestion: `MAX_RETRIES = 42`

2. **Line 48**: Missing input validation
   - Why: Could cause crash on None input
   - Suggestion: Add `if value is None: return default`

### ❗ Critical Issues
1. **Security**: SQL injection vulnerability at line 67
   - Use parameterized queries instead of string formatting

Common Issues to Flag

• Magic numbers without explanation
• Missing error handling
• Security vulnerabilities (injection, XSS, etc.)
• Performance anti-patterns (N+1 queries, etc.)
• Missing tests for critical paths
• Overly complex logic
• Code duplication