Audits code changes, pull requests, and patches to ensure high standards for correctness, security, and performance.
The review-code skill automates the rigorous assessment of code submissions by analyzing logic, security vulnerabilities, and maintenance debt across a standardized five-dimension quality gate. It is specifically designed to identify common AI-generated pitfalls—such as hallucinated API calls or over-engineered logic—while providing actionable, severity-ranked findings rather than generic advice. By integrating structured checks for authentication, data integrity, and validation evidence, it ensures that every pull request meets professional production standards before merging.
主要功能
01Explicit verification for high-risk areas including authentication and data mutations
02Multi-dimensional quality gates covering correctness, security, performance, maintainability, and validation
036 GitHub stars
04Severity-ranked findings (Critical, Important, Suggestion) with executable fix recommendations
05Standardized three-part output structure: Findings, Open Questions, and Summary
06Specialized detection for AI-specific hallucinations and over-engineered logic patterns
使用场景
01Validating AI-generated code snippets to ensure they don't use non-existent APIs or bloated logic
02Automating the first pass of a pull request to catch security flaws and performance bottlenecks
03Enforcing consistent team-wide code quality standards during the CI/CD workflow
