Prompt
Gated Claude x Codex build loop for Claude Code: Codex plans and implements from one persistent thread, Claude approves the plan and reviews the diff. Zero deps, 20-test smoke suite.
npx -y skills add ozzaii/codex-claude-loop --skill promptAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 12 days oldThe repository was created 12 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Send an additional request to the exact Codex session for an implemented codex-claude-loop lane, sequenced behind its writer lock
SKILL.md
1.0 KB, as published. Nobody here has run it
Load the codex-claude-loop skill and source its bundled
lib/codex-claude-loop.sh. Split $ARGUMENTS into the first word as the slug and the
entire remaining text as one request, then call:
cl_prompt <slug> "<additional request>"
Do not call codex exec resume directly. cl_prompt resumes the stored session id,
holds the same writer lock as cl_impl and cl_revise, and waits behind a running lane
instead of starting a competing Codex process in the worktree.
The lane must already have a valid approved plan and a successful implementation marker. The follow-up is a writer phase: it clears any standing review approval, restores the implementation marker only after Codex exits cleanly, and prints Codex's final response. After it succeeds, re-review before release.