Why is the --full-auto flag required for this skill?

The --full-auto flag is critical for headless operation as it auto-approves file reads and command executions that would otherwise cause Codex to hang while waiting for manual user input.

Can I use this skill to review specific files instead of the whole project?

Yes, you can target individual files, multiple specific paths, or entire directories depending on the scope of review you need.

How does the skill ensure dependency information is accurate?

The skill includes a mandatory verification step using WebSearch to confirm the latest version numbers and current best practices before making claims about outdated libraries.

Can this skill be integrated into CI/CD pipelines?

Yes, it is designed to be scriptable and can output results in JSON or Markdown formats, allowing it to act as a quality gate in automated pipelines.

Codex Code Review

Name: Codex Code Review
Author: tyrchen

bytyrchen

•

Security & Testing

Automates comprehensive code reviews and security audits using the OpenAI Codex CLI for deep codebase analysis.

This skill integrates the OpenAI Codex CLI into the Claude environment to perform rigorous, automated analysis of codebases, pull requests, and git diffs. It enables Claude to identify logic errors, security vulnerabilities, performance bottlenecks, and architectural anti-patterns with structured feedback. Designed for both interactive and headless operation, the skill utilizes specific automation flags to provide high-quality code quality reports without manual intervention, making it ideal for pre-commit checks and continuous integration workflows.

Key Features

0111 GitHub stars

02Deep security auditing for OWASP vulnerabilities and hardcoded secrets

03Automated analysis of git diffs, staged changes, and pull requests

04Headless execution mode using the --full-auto flag for automation

05Performance profiling for algorithmic complexity and resource leaks

06Architectural review based on SOLID principles and design patterns

Use Cases

01Auditing large directories or legacy codebases for technical debt and security flaws

02Generating structured feedback for GitHub Pull Requests to streamline peer reviews

03Reviewing uncommitted or staged git changes before making a final commit

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tyrchen/claude-skills codex-code-review

For use in Claude.ai and ChatGPT

Download Skill