Codex Skill for Claude Code plugin
A Claude Code plugin: a /codex skill and a hook that sends every finished plan to Codex CLI for review before you approve it.
- 229 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The hook sends plan texts and code to the external OpenAI service through Codex CLI.
- Installation adds a hook to Claude Code settings, which runs a script on your machine.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/codex-skill-for-claude-codeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The hook sends plan texts and code to the external OpenAI service through Codex CLI.
- Installation adds a hook to Claude Code settings, which runs a script on your machine.
This is third-party code. Review the repository files before installing.
What it does
The plugin adds a second pair of eyes to Claude Code in the form of OpenAI Codex CLI. The hook fires on `ExitPlanMode`: it reads the plan from `tool_response.plan`, passes it to Codex and shows the comments next to the plan, so you approve it with the critique at hand. No model is pinned for the automatic review, the Codex default is used. The `/codex` skill works manually: researching unfamiliar APIs and libraries, verifying complex code patterns, getting an alternative view on architecture. It is a separate request to a second model rather than a continuation of the Claude chat. Plugin install registers both the skill and the hook. Manual install copies `skills/codex/SKILL.md` to `~/.claude/skills/codex/`, the `hooks/plan-review.sh` script to `~/.claude/hooks/` and adds a `PostToolUse` block to `~/.claude/settings.json`.
Who it is for. Claude Code users with an OpenAI key who want an independent check of plans and architecture decisions by a second model.
Good fit when
- A plan for a large change should go to a second model before approval.
- An unfamiliar library is easier to study through Codex first and write code afterwards.
- A tricky piece of code gets a second opinion on edge cases.
- An architecture decision needs an alternative view.
Not a fit when
- Codex CLI is not installed or there is no OpenAI key: the hook and skill depend on them.
- Code and plans cannot be sent to an external provider by company policy.
- Quick edits are enough: reviewing every plan adds a delay of up to 120 seconds.
Example request
/codex review the architecture of this migration plan and name the scenarios where it breaks.Limitations
Codex CLI (`npm install -g @openai/codex`) and a configured OpenAI key are required. Plan texts and code fragments go to OpenAI, and requests spend paid tokens. The hook has a 120-second timeout and works only after `ExitPlanMode`. The list of Codex models in the README changes quickly, so check current names in Codex itself. The license is MIT, the last commit dates from November 2026.
How to disable. Run `claude plugin uninstall codex-skill@codex-skill`. For a manual install delete `~/.claude/skills/codex`, `~/.claude/hooks/plan-review.sh` and the `PostToolUse` block from `settings.json`.
FAQ
How do I install the plugin?
Run `claude plugin marketplace add cathrynlavery/codex-skill`, then `claude plugin install codex-skill@codex-skill`. Both the skill and the hook register on their own.
When does the automatic review fire?
After `ExitPlanMode`, when Claude Code is ready to show the plan. Codex reads it and returns comments before you approve.
Is an OpenAI key required?
Yes, Codex CLI works with your OpenAI key, which is configured before installing the plugin.
How do I trigger a review manually?
With the `/codex` command and a question, for example to study an unfamiliar API or check an architecture choice.
Which Codex model does the hook use?
The one you chose as the Codex default: the hook does not pin a model.
Discussion
No questions or reviews yet.
Related
PluginsOpenAI's official plugin: Codex reviews of your changes and task delegation to Codex from inside a Claude Code session
A Claude Code skill that delegates tasks to OpenAI's Codex CLI for code analysis, review and edits
A Codex plugin that runs Claude Code for reviews, rescue tasks and tracked background jobs without leaving Codex
A Claude Code skill that drives Codex over JSON-RPC: tasks, code reviews, parallel research and messaging between two agents.
Codex Settings skills
Codex CLI Skills & Profiles
A Codex CLI plugin with twelve skills: deep research, pull request review, image generation and session handoff
A Claude Code plugin: Claude does the task, then a stop hook starts an independent Codex review and Claude fixes the findings in the same session
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Codex Skill for Claude Code (this card) | no | 229 | |
| Codex plugin for Claude Code | no | 35,088 | Codex plugin for Claude Code delegates tasks and code review to Codex, while this plugin hangs a plan review on the ExitPlanMode hook. |
| Claude-Codex Review Skill | no | 5 | Claude-Codex Review Skill stays a manual review, while here the plan goes to review automatically. |
README in short
A plugin with an ExitPlanMode hook and a manual /codex skill: Claude Code gets an independent plan review from Codex CLI.