codex

Review code diffs with Codex CLI in review, challenge, or consult modes.

Updated Apr 6, 2025

Search Tags:#codex #diff #developer-tools #code-review #adversarial #ai-assistance

Ready to run

npx skills add https://github.com/yolopunk/jedi --skill codex-yolopunk

💬 Prompt for ChatGPT / Claude Chat▼

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Independently assesses code changes using Codex CLI, providing a brutally honest second opinion to catch issues that slip through standard reviews.

Core Features & Use Cases

Three modes: code review with a pass/fail gate, adversarial challenge to stress-test changes, and consult prompts with session continuity.
Deterministic gating: outputs a gate verdict and token usage metrics to inform decision-making.
Session-aware conversations: retains context across interactions for follow-ups and long-running discussions.

Quick Start

Run codex review against your current branch to obtain an independent assessment of the 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 git diff before merging?▼

You can get an independent AI code review by running the codex review mode on your branch diff against the base branch. It assesses code changes and outputs a pass/fail gate verdict along with token usage metrics to inform merge decisions.

What is adversarial code review and how does it stress-test branch changes?▼

Adversarial code review uses a challenge mode to stress-test your branch changes by prompting an AI assistant to find edge cases and flaws. It delivers a brutally honest second opinion to catch issues that slip through standard reviews.

Can I use an AI assistant for code review with session persistence for follow-up questions?▼

Yes, you can use the consult mode to conduct AI code reviews with session persistence. It retains conversational context across interactions, enabling follow-up questions and long-running discussions about your code changes.

Does this code review tool handle automation timeouts and errors for CI pipelines?▼

Yes, the code review tool is designed for automation with robust error handling and timeout management. It provides gate-style output suitable for logging and CI pipelines, ensuring reliable automated code assessments.