code-review

Invoke the code-review skill after significant code changes. It analyzes six aspects including architecture, security, and performance, then returns prioritized issues with root cause analysis and working code solutions for each finding.

This review covers six areas: architecture and design, code quality, security and dependencies, performance and scalability, testing quality, and documentation and API design. Multiple review instances can run in parallel for full coverage.

Issues are classified into four tiers: Critical for security vulnerabilities and data loss risks, High for hot-path performance and broken error handling, Medium for maintainability concerns, and Low for style inconsistencies and minor optimizations.

Yes. Before reviewing, it reads project documentation like README and CONTRIBUTING files, detects patterns such as Result types, dependency injection, and state management, then verifies new code follows the discovered conventions.

Yes. It uses grep and git history analysis to find duplicated logic across files, identify high-churn files needing refactoring, check test coverage for reviewed components, and verify all call sites are handled after changes.