codex

Name: codex
Availability: InStock
Author: kakarottoooo

Review code diffs via OpenAI Codex CLI in three modes.

Updated Mar 17, 2026

Search Tags:#automation #cli #codex #engineering #openai #code-review #adversarial

Ready to run

npx skills add https://github.com/Kakarottoooo/onegent --skill codex-kakarottoooo

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/Kakarottoooo/onegent/tree/main/.agents/skills/gstack/codex
Command: npx skills add https://github.com/Kakarottoooo/onegent --skill codex-kakarottoooo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Codex is wrapped to provide an independent, brutally honest second opinion on code changes by leveraging the OpenAI Codex CLI, helping teams validate diffs, catch edge cases, and sanity-check decisions.

Core Features & Use Cases

Review mode: run Codex code review against the current branch diff, with gating to accept or reject changes.
Challenge mode: generate adversarial prompts to probe for edge cases, security holes, or failure modes.
Consult mode: ask Codex questions about the codebase or project prompts and receive continuity across questions.

Quick Start

Use /codex to ask for a review, challenge, or consultation on your code, e.g., "/codex review ..." or "/codex challenge ..."

Frequently Asked Questions about codex

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

FAQPage Schema

How do I get an independent code review on my current git branch diff?▼

You can get an independent code review by running the Codex CLI against your current branch diff. The review mode evaluates your changes and applies gating to accept or reject them based on the analysis.

How do I probe my code changes for edge cases and security holes?▼

You can probe for edge cases and security holes by using challenge mode, which generates adversarial prompts to actively test your code changes for failure modes and weaknesses.

Can I ask questions about my codebase and maintain context across queries?▼

Yes, you can ask questions about your codebase while maintaining context across queries by using consult mode, which leverages the Codex CLI to provide continuous answers about your project.

Do I need to install the OpenAI Codex CLI to use this automation workflow?▼

Yes, you must install the OpenAI Codex CLI locally to use this workflow. The Skill operates by wrapping the CLI and relies on your existing git state and prompts to determine the appropriate mode.

What is the best way to sanity-check architectural decisions in my codebase?▼

The best way to sanity-check decisions is using consult mode, which asks Codex questions about your project prompts and codebase to validate architectural choices and catch potential issues.

When should I use adversarial prompts instead of a standard code review?▼

You should use adversarial prompts when a standard review is insufficient and you need to aggressively probe for specific failure modes or security vulnerabilities before merging changes.