codex

Wraps Codex CLI to review, challenge, and consult on diff-based code changes.

43|7|Updated Mar 13, 2026
One-click install
npx skills add https://github.com/lucas-flatwhite/gstack-ko --skill codex-lucas-flatwhite
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/lucas-flatwhite/gstack-ko/tree/main/codex
Command: npx skills add https://github.com/lucas-flatwhite/gstack-ko --skill codex-lucas-flatwhite

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This tool provides an independent, brutally honest second opinion on code changes by wrapping the OpenAI Codex CLI to review diffs, challenge assumptions, and answer questions with session continuity.

Core Features & Use Cases

  • Review mode: code review against the current branch diff with a gate-style verdict.
  • Challenge mode: adversarial probing to uncover edge cases, security risks, and possible failure modes.
  • Consult mode: ask Codex questions with session continuity for follow-ups and clarifications.

Quick Start

Run /codex review, /codex challenge, or /codex consult to get an independent opinion on your code changes.

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 AI code review on my current branch diff?

To get an independent AI code review on your branch diff, you can use the review mode, which evaluates your uncommitted changes against the base branch and provides a gate-style verdict on code quality.

What is adversarial code challenge and how does it find edge cases?

Adversarial code challenge is a probing mechanism that questions your assumptions to uncover security risks, edge cases, and failure modes, providing an independent analysis of possible design flaws in your changes.

Do I need the Codex CLI installed to review base-branch diffs?

Yes, you need the Codex CLI installed locally, as this tool wraps the CLI to process base-branch diffs, execute adversarial challenges, and manage session-based consultations for follow-up questions.

Can I ask follow-up questions after an initial code review verdict?

Yes, you can ask follow-up questions using the consult mode, which maintains session continuity to provide clarifications and answers to your questions after an initial review or challenge.

What is the best way to catch design flaws before merging code changes?

The best way to catch design flaws before merging is to run an adversarial challenge against your base-branch diff, which probes for failure modes and provides a brutally honest second opinion on your logic.

Code review tools not catching edge cases in my diffs, what are the limitations?

Standard code review tools may miss edge cases because they lack adversarial probing; wrapping the Codex CLI with a challenge mode directly questions assumptions to uncover security risks and failure modes.