Code Reviewer Agent

The Code Reviewer Agent provides automated, AI-powered code reviews on every pull request. It analyzes code changes, identifies issues, and suggests improvements.

What It Does

The Code Reviewer Agent:

• Analyzes diffs - Reviews only the changed code, not the entire codebase
• Identifies issues - Catches bugs, logic errors, and anti-patterns
• Suggests improvements - Recommends better approaches and optimizations
• Enforces standards - Checks for style guide compliance
• Documents concerns - Explains why each issue matters

Supported Languages

| Language | Support Level | |----------|--------------| | TypeScript/JavaScript | Full | | Python | Full | | Go | Full | | Rust | Full | | Java | Full | | C/C++ | Partial | | Ruby | Partial |

Configuration Options

agents:
  - name: code-reviewer
    template: code-reviewer
    config:
      # Review thoroughness: quick, balanced, thorough
      review_style: balanced

      # Minimum severity to report: info, warning, error
      min_severity: warning

      # Auto-approve PRs with no issues
      auto_approve: false

      # Focus on specific file patterns
      include_patterns:
        - "src/**/*.ts"
        - "lib/**/*.py"

      # Ignore certain files
      exclude_patterns:
        - "**/*.test.ts"
        - "**/__mocks__/**"

      # Review aspects to focus on
      focus_areas:
        - security
        - performance
        - maintainability
        - testing

Review Output

The agent posts a structured review comment:

Summary

A brief overview of the changes and overall assessment.

Issues Found

Each issue includes:

• Severity - Error, Warning, or Info
• Location - File and line number
• Description - What the issue is
• Suggestion - How to fix it
• Code snippet - Suggested replacement code

Example Review

## Code Review Summary

This PR adds user authentication to the API. Overall, the implementation
looks good with a few suggestions for improvement.

### Issues Found

#### 1. Potential SQL Injection (Error)
📍 `src/auth/login.ts:45`

The user input is directly interpolated into the SQL query without
sanitization.

**Current:**
\`\`\`typescript
const query = `SELECT * FROM users WHERE email = '${email}'`;
\`\`\`

**Suggested:**
\`\`\`typescript
const query = 'SELECT * FROM users WHERE email = $1';
const result = await db.query(query, [email]);
\`\`\`

#### 2. Missing Error Handling (Warning)
📍 `src/auth/login.ts:52`

The async operation doesn't handle potential errors...

Best Practices

1. Start with balanced mode - Adjust after seeing initial reviews
2. Exclude test files - Unless you want test reviews too
3. Set minimum severity - Start with "warning" to reduce noise
4. Review the reviews - Agents learn from your feedback

Integration with Approval Workflows

Combine with approval workflows:

approval_workflow:
  require_review: true
  block_on_errors: true
  allow_override: true