You can now trigger Codex from Claude Code! Here's how
AlphaSignal
The most read source of technical news in AI. We help you stay up to date with the latest news, research, models.
A new Claude Code plugin, codex-plugin-cc, lets you run Codex directly inside Claude Code.
The idea that you can pull in a second AI coding agent from within the same terminal, from a competing company no less, is something we haven’t seen before.
What Is codex-plugin-cc?
Once installed, a set of /codex: slash commands appear in your terminal session. You can invoke Codex for code reviews, adversarial security audits, bug investigations, and full task delegation without switching tools.
It requires either a ChatGPT subscription (including the Free tier) or an OpenAI API key, plus Node.js 18.18 or later. The plugin wraps your local Codex CLI and Codex app server, so it picks up whatever authentication, configuration, environment variables, and MCP setup you already have.
Here are the installation steps:
To install the codex plugin in Claude, you need to add the plugin first into the marketplace by running this command:
/plugin marketplace add openai/codex-plugin-cc
You can find more details of the plugin source code here: https://github.com/openai/codex-plugin-cc
Next, install the plugin by running the command:
/plugin install codex@openai-codex
You will be asked to which scope you want the plugin to be accessible.
Choose the one you prefer and run the “/codex:setup” command.
In case the command is not recognized, do a “/reload-plugins” to refresh the list and try again. You should see the following message:
Select “Install Codex” and wait for the setup to complete. If all looks good, you should see a “Codex is ready” status like below.
Codex commands
When you type “/codex” this will bring up the following commands:
Use the /codex:rescue command when you want Codex to:
Recommended by LinkedIn
It also supports --wait, --resume, and more. Here are some examples:
/codex:rescue investigate why the tests started failing
/codex:rescue fix the failing test with the smallest safe patch
/codex:rescue --resume apply the top fix from the last run
/codex:rescue --model gpt-5.4-mini --effort medium investigate the flaky integration test
/codex:rescue --model spark fix the issue quickly
/codex:rescue --background investigate the regression
How it works
The plugin delegates through the local Codex CLI and Codex app server. So it uses the same local auth, config, environment, and MCP setup you already have with Codex.
It is basically just Codex being invoked in Claude Code.
Here’s an example of me running Codex review. status, cancel, and results inside Claude Code terminal.
You can also enable a review gate:
bash/codex:setup --enable-review-gate
That can stop Claude Code from exiting before a Codex review has run.
Useful, but something to use carefully. It can create a long Claude/Codex loop and burn through usage limits quickly.
What It Tells Us About the AI Coding Market
When the products are this similar, the differentiators shift to plugin availability, pricing, and developer culture.
Claude Code leads in VS Code marketplace installs (5.2 million vs. 4.9 million for Codex) and costs half the price at $100/month for Max compared to Codex’s $200 entry point. But Codex is open-source, backed by a Free tier, and now building cross-platform bridges that Anthropic hasn’t matched.
codex-plugin-cc is a small piece of a larger trend. Developers don’t want to pick one agent and abandon the rest, they want to compose workflows from the best tools available regardless of who built them. OpenAI is betting that showing up inside a competitor’s terminal is a better growth strategy than walling off the garden.
What this looks like in practice: use Claude Code’s Opus 4.6 for architecture planning and complex feature work, then hand off targeted bug fixes to Codex’s gpt-5.4-mini when speed matters more than reasoning depth. Or let Claude Code write the implementation and have Codex run an adversarial review before every merge.
What do you think about this move? Is OpenAI playing nice, or playing chess?
References:
Join 250k+ developers staying ahead in AI. We curate the latest models, repos, and research — so you don’t miss what matters: AlphaSignal.ai
i just want to know how you do this shiny thing with your logo
My two cents: Claude should design the building. Codex should lay the bricks. Claude should inspect the building. That is the correct way to use both together.