jeff-yiii/claude-code-review

Multi-Agent Code Review System: a Claude Code–based tool for collaborative code review and automated fixes. Intended for developers and teams who want automated pre-commit review, in-session status, and security checks integrated into a repo. Key parts: - Skills layer: three agents — /code-review (Explorer + Reviewer two-stage pipeline), /auto-fix (automatic low-risk fixes), /security-scan (deep security auditor). - Hooks: PreCommit (prompts to run /code-review), SessionStart (show status). - Knowledge: CLAUDE.md (coding rules, anti-patterns, security rules) and persistent memory files. Core features: - /code-review scans git changes, assesses impact, and checks per-file across SPEC/SAFETY/BUG/STYLE, producing structured reports with file paths, line numbers, and fix suggestions. - /auto-fix applies safe fixes (naming, magic numbers, import order, var→const, ==→===) and generates plans for high-risk issues. - /security-scan detects SQLi, XSS, command/path injection, hard-coded secrets, auth flaws, sensitive data leaks, and runs dependency audits. Severity levels: CRITICAL/HIGH/MEDIUM/LOW with recommended actions. Adaptable to projects by adding CLAUDE.md and the .claude directory and tailoring rules.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.