codex

Run diff-based code reviews, adversarial challenges, and consult prompts via Codex CLI.

1|Updated Feb 25, 2026

Search Tags:#security #cli #codex #ai-assistant #code-review #adversarial

Ready to run

npx skills add https://github.com/alexandervashurin/semaphore --skill codex-alexandervashurin

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/alexandervashurin/semaphore/tree/main/.claude/skills/codex
Command: npx skills add https://github.com/alexandervashurin/semaphore --skill codex-alexandervashurin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

OpenAI Codex CLI wrapper that provides three modes for code evaluation: code review with a diff-based gate, an adversarial challenge to uncover edge cases and vulnerabilities, and a consult mode with session-continuity prompts. It delivers independent, rigorous feedback to accelerate safer, higher-quality code shipping.

Core Features & Use Cases

Code review with gate: Perform a diff-based review that can pass or fail a change before merging.
Adversarial challenge: Probe changes for edge cases, potential failures, and security considerations.
Consult prompts: Engage Codex for targeted questions or explanations with session continuity for follow-ups.

Quick Start

Run the codex CLI in your desired mode (review, challenge, or consult) with your project changes to receive a detailed Codex analysis and guidance.

Frequently Asked Questions about codex

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I automate code review to pass or fail a pull request before merging?▼

To automate code review, you run the Codex CLI in review mode against your project changes. It performs a diff-based pass/fail gate evaluation that provides independent, rigorous feedback to ensure safer, higher-quality code shipping before merging.

What is adversarial code review and how does it find security vulnerabilities?▼

Adversarial code review probes changes for edge cases, potential failures, and security considerations. The challenge mode actively attempts to break code changes to uncover hidden vulnerabilities and architectural flaws before deployment.

Do I need OpenAI Codex CLI installed and authenticated to run adversarial challenges?▼

Yes, the Codex CLI must be installed and authenticated. This Skill operates as a wrapper for the Codex CLI, requiring proper setup and authentication to execute code reviews, adversarial challenges, and consult prompts.

Can I use consult mode for architectural discussions with session continuity?▼

Yes, you can use consult mode for architectural discussions with session continuity. It engages Codex for targeted questions or explanations while maintaining session context for follow-up inquiries during code evaluation.

What's the best way to probe code changes for edge cases and potential failures?▼

The best way to probe code changes for edge cases is using an adversarial challenge mode. This approach rigorously tests your changes by attempting to break them, uncovering potential failures and security considerations before deployment.