OpenAI has released a Codex plugin for Claude Code, allowing users to call Codex from within Claude Code to perform code reviews and task delegation.
OpenAI officially released a plugin on March 31, 2026, that allows users to directly utilize OpenAI Codex within the Claude Code workflow for code reviews and task delegation. By installing this plugin, developers can receive double-checking from different AI agents without leaving Claude Code.
openai/codex-plugin-cc: Use Codex from Claude Code to review code or delegate tasks.
This Codex plugin is not a standalone runtime but works in conjunction with a locally installed Codex CLI or application server. Therefore, existing Codex authentication status and configuration files such as ~/.codex/config.toml are retained. Use requires a ChatGPT subscription (including the free version) or an OpenAI API key, and a Node.js 18.18 or later environment is required. Installation can be done by adding the plugin from the Claude Code Marketplace and running the dedicated setup command.
Starting today you can use Codex in Claude Code ????
— Vaibhav (VB) Srivastav (@reach_vb) March 30, 2026
/plugin marketplace add openai/codex-plugin-cc
Try it out today with:
/codex:review for a normal read-only Codex review
/codex:adversarial-review for a steerable challenge review
/codex:rescue to let codex rescue your code… https://t.co/VdVqstzwb4 pic.twitter.com/aA5HGc3O4C
The main features of the Codex plugin include a standard code review function that allows you to view current changes in read-only mode, an adversarial review function that rigorously examines implementation design, trade-offs, and potential risks, and a rescue function that allows you to directly request bug investigation and fixes from Codex.
The code review feature, /codex:review, is performed on uncommitted changes and comparisons with a specific branch, providing the same quality as running /review directly in Codex.
You can now trigger @OpenAI Codex from Claude Code!
— Derrick Choi (@derrickcchoi) March 30, 2026
My recommendation is to use Codex to review Claude Code changes ????
Install the plugin here: https://t.co/P7HZqAstpC pic.twitter.com/b71P4tgZ8U
On the other hand, the adversarial review feature is particularly effective when hidden prerequisites pose a risk, such as changes to authentication, infrastructure scripts, or large-scale refactoring, and it is also possible to specify areas of focus and direct the review accordingly.
The rescue function allows you to delegate tasks by specifying a particular model, such as gpt-5.4-mini, or by adjusting the inference strength. These tasks are recommended to run in the background, and you can check their progress, view results, and pause them using dedicated commands.
Furthermore, enabling the review gate feature blocks the termination of Claude Code until the Codex review is complete, prompting Claude to address any issues found. However, this feature can rapidly consume available API tokens, so caution is advised.
Related Posts:
