codex-plugin-cc

This plugin connects Claude Code to Codex, letting you run code reviews or hand off tasks to Codex without switching tools. It's built for developers who already work in Claude Code and want Codex's review and task-delegation capabilities available as slash commands, without leaving their session.

Under the hood it wraps the Codex app server and uses your locally installed codex binary, so it reuses your existing Codex authentication, configuration, and repository checkout. There's no separate account or runtime to manage.

Key features

• /codex:review: runs a read-only Codex review of uncommitted changes or a branch diff against a base ref, with --wait or --background options.
• /codex:adversarial-review: a steerable review that challenges design decisions, tradeoffs, and hidden assumptions instead of just flagging code issues.
• /codex:rescue: delegates a task (investigate a bug, try a fix, resume a prior thread) to Codex through a dedicated subagent, with model and effort overrides.
• /codex:transfer: converts the current Claude Code session into a persistent Codex thread you can resume in the Codex App or TUI.
• /codex:status, /codex:result, /codex:cancel: check progress, retrieve final output, or cancel background Codex jobs tied to the current repo.
• /codex:setup: checks whether Codex is installed and authenticated, and can install it via npm if missing.
• Optional review gate: a Stop hook that runs a targeted Codex review on Claude's response and blocks the stop if issues are found (opt-in, since it can loop and consume usage quickly).
• Config reuse: picks up your existing user-level or project-level .codex/config.toml, including model and reasoning-effort defaults.

Ideal use cases

Use this plugin if you're already running Claude Code as your primary workflow and want a second opinion or a delegation path to Codex without context-switching. It fits teams doing pre-merge reviews (/codex:review --base main), pressure-testing design decisions before shipping (/codex:adversarial-review), or offloading a stuck bug investigation to a background Codex job while continuing other work in Claude Code.

It's also useful when you want to start a long-running task in the background and check back later with /codex:status and /codex:result, rather than blocking on a synchronous review.

It's not a fit if you don't already use Claude Code, since this is a Claude Code plugin, not a standalone tool. It's also not useful if you have no Codex access (ChatGPT subscription or OpenAI API key) or can't install Node.js 18.18+, both of which are hard requirements. If you want a fully autonomous review loop with no oversight, be cautious: the review gate explicitly warns it can create a long Claude/Codex loop that drains usage limits if left unmonitored.

Installation

Requirements: a ChatGPT subscription (including Free) or an OpenAI API key, and Node.js 18.18 or later.

Add the plugin marketplace inside Claude Code:

/plugin marketplace add openai/codex-plugin-cc

Install the plugin:

/plugin install codex@openai-codex

Reload plugins:

/reload-plugins

Run setup to verify Codex is ready (it can offer to install Codex via npm if missing):

/codex:setup

To install Codex manually instead:

npm install -g @openai/codex

If Codex is installed but not logged in:

!codex login

A simple first run to confirm everything works:

/codex:review --background
/codex:status
/codex:result