# codex-plugin-cc

> Run Codex code reviews and hand off tasks without leaving Claude Code

- Repository: https://github.com/openai/codex-plugin-cc
- Category: Agent skills & plugins
- Stars: 34,500 · Forks: 2,381 · Language: JavaScript · License: Apache-2.0
- Experience: Beginner friendly
- Page: https://ir.adyog.com/repos/openai/codex-plugin-cc/

If you work in Claude Code but also have access to OpenAI's Codex, switching between the two is clumsy. This plugin adds slash commands to Claude Code so you can ask Codex to review your changes or take on a task, then read the results in the same place.

It does not bring its own runtime. It talks to the Codex CLI and Codex app server already installed on your machine, so it uses your existing Codex login, config file, and repository checkout. Long jobs can run in the background, and you can check on them or cancel them with other commands.

There are two kinds of review. The normal review is read-only and matches what /review does inside Codex. The adversarial review accepts extra focus text and challenges design choices and risks. An optional review gate can also run a Codex review before Claude stops, but the README warns it can burn through usage limits quickly.

## What it does

- **Read-only code review**: /codex:review reviews your uncommitted changes or a branch compared to a base such as main. It can run in the foreground or the background.
- **Adversarial review**: /codex:adversarial-review takes focus text and questions assumptions, tradeoffs, and failure modes like race conditions or data loss. It does not change code.
- **Delegate tasks to Codex**: /codex:rescue hands a bug investigation or fix to a Codex subagent. You can pick a model and effort, and resume the previous thread.
- **Background job control**: /codex:status, /codex:result and /codex:cancel let you follow, read, or stop long-running Codex jobs. Results include a session ID for codex resume.
- **Session transfer and review gate**: /codex:transfer turns your Claude Code session into a Codex thread you can continue with codex resume. An optional Stop-hook review gate blocks Claude from stopping if Codex finds issues.

## What you could build

- Get a second-opinion review of your uncommitted changes before committing.
- Challenge a caching or retry design with an adversarial review focused on race conditions.
- Hand a failing CI build or flaky test to Codex while you keep working in Claude Code.
- Continue a debugging conversation started in Claude Code inside the Codex app or terminal.
- Run a long review in the background and check its status and result later.

## Who it's for

Claude Code users who also have a ChatGPT account or OpenAI API key and are comfortable with slash commands and Node.js 18.18 or later.

## Get started

Add the marketplace and install the plugin in Claude Code, then reload plugins and run /codex:setup to check that Codex is installed and logged in. A simple first run is /codex:review --background, then /codex:status and /codex:result.

```
/plugin marketplace add openai/codex-plugin-cc
```
```
/plugin install codex@openai-codex
```
```
/codex:setup
```

Tags: claude-code, codex, code-review, plugin, openai