codex

Automate Codex-driven code review and adversarial probing via CLI.

2|Updated May 8, 2026
One-click install
npx skills add https://github.com/xotong/claude-marketplace --skill codex-xotong
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/xotong/claude-marketplace/tree/main/plugins/gstack/skills/codex
Command: npx skills add https://github.com/xotong/claude-marketplace --skill codex-xotong

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Codex CLI wrapper provides an automated, structured code review and adversarial probing workflow, enabling developers to obtain a second opinion and robust feedback without leaving the terminal.

Core Features & Use Cases

  • OpenAI Codex-powered code review in three modes: review, challenge, and consult, with session continuity for follow-ups.
  • Deterministic decision gates and reproducible prompts to catch edge cases and potential vulnerabilities.
  • Suitable for diff reviews, security-checks, and interactive coding sessions where a quick, expert assessment is required.

Quick Start

Run codex review to start an expert, session-based code-review workflow.

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 for diffs directly from the CLI?

Automate code review for diffs by running the codex review command, which initializes a session-based workflow to analyze code changes and provide expert feedback directly in your terminal.

What is adversarial probing in AI-driven code review?

Adversarial probing in code review is a challenge mode that applies deterministic decision gates and reproducible prompts to actively catch edge cases and potential vulnerabilities in your code.

Do I need OpenAI Codex access to run CLI-based code reviews?

Yes, you need OpenAI Codex access to run these CLI-based code reviews, as the wrapper relies on Codex to power the expert assessment, security checks, and interactive consultations.

Can I maintain session continuity for follow-up questions during a code review?

Yes, you can maintain session continuity for follow-up questions during interactive coding sessions, allowing you to consult the AI with context-aware inquiries without restarting the workflow.

What's the best way to perform security-sensitive code checks without leaving the terminal?

The best way to perform security-sensitive checks without leaving the terminal is using the CLI wrapper's gated context loading, which applies reproducible prompts to catch potential vulnerabilities in your diffs.

When should I use consult mode versus challenge mode for code analysis?

Use consult mode for interactive coding sessions requiring expert assessment and follow-ups, while challenge mode is for adversarial probing to catch edge cases and vulnerabilities through deterministic decision gates.