codex

Get a second opinion on code changes by running an independent diff review with a pass or fail gate. The Codex CLI wrapper evaluates your changes and provides a brutally honest assessment of whether they meet quality standards.

Adversarial code testing is a mode that actively tries to break your code changes. The challenge mode attacks your logic to uncover edge cases and vulnerabilities, providing a rigorous stress test before deployment.

Yes, the Codex CLI must be installed to use this code review skill. The wrapper invokes the CLI directly with appropriate prompts for review, challenge, and consult modes.

Yes, you can ask follow-up questions interactively using the consult mode. It provides session continuity, allowing you to maintain context across multiple prompts when consulting about your project.

An independent AI code review goes beyond standard linters by providing a brutally honest second opinion on logic and architecture. It evaluates diffs contextually, offering adversarial testing and interactive consultation rather than just syntax checks.

Use adversarial testing instead of a standard code review when you need to stress-test logic against edge cases. The challenge mode actively attempts to break your changes, making it ideal for critical paths requiring robust validation.