explaingit

openai/codex-plugin-cc

★ 18,996JavaScriptAudience · developerComplexity · 2/5MaintainedLicenseSetup · moderate

TLDR

A Claude Code plugin that lets you call OpenAI's Codex AI agent for code reviews, bug fixes, and investigative tasks without leaving your editor.

Mindmap

mindmap
  root((repo))
    What it does
      Code reviews
      Bug investigation
      Design challenges
      Background tasks
    Key commands
      slash codex review
      slash codex rescue
      Job management
    Setup requirements
      ChatGPT subscription
      Node.js 18.18
      Claude Code editor
    Use cases
      Peer review automation
      Test failure debugging
      Design validation

Things people build with this

USE CASE 1

Run automated code reviews on uncommitted changes or branches to catch issues before merging.

USE CASE 2

Hand off bug investigations and repairs to a background Codex agent while you continue other work.

USE CASE 3

Challenge your design decisions with adversarial reviews that examine tradeoffs and failure modes.

USE CASE 4

Set up a review gate that forces Claude to address Codex findings before responding.

Tech stack

JavaScriptNode.jsOpenAI APIClaude Code

Getting it running

Difficulty · moderate Time to first run · 30min

Requires OpenAI API key and Claude Code editor setup with plugin installation.

Use freely for any purpose including commercial. Keep the notice and disclose changes to the patent grant.

In plain English

The Codex plugin for Claude Code is a plugin that lets Claude Code users invoke Codex, a separate AI coding agent from OpenAI, from within their existing Claude Code workflow. It is designed for developers who already use Claude Code and want to occasionally hand tasks or reviews off to Codex without switching to a different tool. Once installed, the plugin adds several slash commands. /codex:review runs a read-only code review on your current uncommitted changes or on a branch compared to a base branch. /codex:adversarial-review runs a more challenging review that questions design decisions and examines tradeoffs, failure modes, and whether alternative approaches would have been safer or simpler. /codex:rescue hands an investigative or repair task, such as finding why tests are failing or attempting a bug fix, to a Codex subagent that runs in the background. /codex:status, /codex:result, and /codex:cancel let you monitor and manage background Codex jobs. An optional review gate can be enabled that automatically runs a Codex review before Claude Code finishes responding; if the review finds problems, Claude must address them first. The readme warns this can create long-running loops that consume usage limits quickly. The plugin requires a ChatGPT subscription or OpenAI API key, Node.js 18.18 or later, and installation through the Claude Code plugin marketplace.

Copy-paste prompts

Prompt 1
Install the Codex plugin for Claude Code and run /codex:review on my current branch to get a code review.
Prompt 2
Use /codex:rescue to investigate why my test suite is failing and attempt a fix in the background.
Prompt 3
Enable the optional review gate so Codex automatically reviews my code before Claude Code finishes responding.
Prompt 4
Run /codex:adversarial-review to challenge the design decisions in my recent changes and suggest safer alternatives.
Prompt 5
Check the status of my background Codex job with /codex:status and retrieve results with /codex:result.
Open on GitHub → Explain another repo

Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.