Automates multi-dimensional code, plan, and scope reviews using specialized AI experts and static analysis tools.
Harness Review is a comprehensive quality assurance skill for Claude Code that automates the critical review phase of the autonomous development lifecycle. It intelligently switches between Code, Plan, and Scope review modes based on your recent activity, utilizing a 'Quality Gate' to identify high-priority areas such as security vulnerabilities, performance bottlenecks, and accessibility gaps. By integrating with LSP and AST-Grep, it performs deep structural analysis and impact assessments, and can even trigger parallel sub-agents for large-scale changes to ensure production-grade quality before any code is committed.
主要功能
01160 GitHub stars
02AST-Grep integration for structural code smell detection and pattern matching.
03Context-aware review mode detection for code, project plans, and task scope.
04Security-focused auditing covering OWASP Top 10, authentication, and input validation.
05LSP-based impact analysis to identify breaking changes across the entire codebase.
06Parallel expert execution for multi-faceted reviews of large-scale pull requests.
使用场景
01Performing a comprehensive security and quality audit before submitting a Pull Request.
02Automated impact analysis when refactoring core functions or modifying shared type definitions.
03Validating development plans and feature scope to prevent scope creep and identify hidden dependencies.
