codex-code-reviewer

Automated code review systematically scans recently modified files across five dimensions—quality, security, performance, architecture, and documentation—identifying issues like SQL injection risks and suggesting fixes iteratively until code meets standards.

Yes, the Skill applies fixes iteratively with optional approval: in automated mode it autonomously applies severity-based corrections, while in final validation mode it enforces dual-tool verification before completion, logging all decisions.

It detects quality violations, security vulnerabilities like SQL injection and N+1 queries, performance bottlenecks, architectural inconsistencies, and documentation gaps across your codebase in a single review cycle.

The Skill identifies issues, proposes corrections, applies them based on your automation settings, then re-validates the updated code, repeating cycles until all detected problems are resolved and quality gates pass.

Use automated review in development workflows after code generation, post-commit, or across diverse projects to ensure consistency; manual review remains valuable for architectural decisions that require human judgment.

Yes, the Skill operates on recently modified code files across diverse project types, reading automation mode and coverage targets from context to adapt its validation strategy to your workflow.