@reach_vb - OpenAI Codex Plugin Claude

Open original

@reach_vb on X OpenAI has released a plugin that lets you run Codex inside Claude Code, so you can get a second opinion from a different AI agent without switching tools. It offers a standard review, a more skeptical adversarial review, and a way to hand off whole tasks to Codex.

  • The plugin calls the local Codex CLI and app server, so it reuses your existing Codex auth, config, and MCP setup rather than running a separate runtime.
  • Commands include /codex:review for a standard read-only pass, /codex:adversarial-review for a challenge-style critique, and /codex:rescue to take over a stalled task; longer jobs can run in the background.
  • An optional review gate can block Claude Code from exiting until Codex has reviewed, which the author warns can create long back-and-forth loops and burn through usage limits quickly.

OpenAI has released a plugin that lets you call Codex from inside Claude Code, so you can get a second opinion from a different model without switching tools. Setup needs Node.js 18.18 or later and either a ChatGPT subscription (Free tier works) or an OpenAI API key. The plugin is a thin wrapper: it calls your local Codex CLI and Codex app server, so it reuses whatever auth, config, environment variables, and MCP servers you already have set up for Codex. It isn't a separate runtime, which is why it stays lightweight. There are three main commands. `/codex:review` runs a standard read-only review. `/codex:adversarial-review` is the more interesting one: it's steerable and designed to challenge the implementation rather than just inspect it. `/codex:rescue` hands a task to Codex outright, useful when a thread in Claude Code has stalled. Longer jobs can go to the background and be managed with `/codex:status`, `/codex:result`, and `/codex:cancel`. The author's suggested default is to run `/codex:review` on everything and save adversarial review for high-stakes work, specifically migrations, auth changes, infrastructure scripts, and refactors, where the real risk is usually an unstated assumption rather than a syntax error. The article also mentions an optional review gate that can block Claude Code from exiting until a Codex review has run. The author flags this as something to use carefully, since it can set off a long back-and-forth between the two agents and chew through usage limits quickly. Install instructions, the repo link (github.com/openai/codex-plugin-cc), and the first-run steps, starting with `/codex:setup`, are all in the post.

Ask about this article