claudex-loop
Lets Claude Code and Codex review each other's plans and code, or routes a task to the better fit.
- Ecosystem
- Agent Skills
- Skill category
- Coding
- Creator / organization
- chaseai-yt
What it is
claudex-loop, published by GitHub user chaseai-yt, is described by the author as Skills for working across Claude Code and Codex. It can route a task to whichever side fits better, or have one side write a plan or code while the other reviews it. The full review loop needs both command-line tools; the route part works on its own. With two models taking turns to check the work, mistakes one model misses have a second chance of being caught.
Who it's for:Developers who have both Claude Code and Codex and want them to check each other. Readers who don't write code won't need it.
Before you start: Install through a Marketplace or by copying the Skills. You need Python 3.10 or later and both command-line tools logged in. Documentation is mainly in English. Your code and questions go to both Claude and Codex, using subscription or API allowance on both sides. MIT license; some content comes from Matt Pocock, with third-party notices included.
Based on the project's public documentation. Not tested by AI Inspiration World.