REPO PACKAGE
openai codex-plugin-cc
OFFICIAL3skills
1ready to use
LABSCO SUMMARY
The plugin adds eight commands under /codex:* — review, adversarial-review, rescue, transfer, status, result, cancel, and setup — that hand work to your local Codex install rather than running any model call of their own. /codex:review and /codex:adversarial-review read your working tree and produce a critique, the second one steerable toward a specific worry like auth or race conditions; /codex:rescue delegates a fix or investigation to Codex in the background; /codex:transfer exports the current Claude Code session so you can pick the same context up inside Codex.
It only works if Codex is already something you use. It shells out to the codex binary and your existing login, so it needs Node.js 18.18 or later, a ChatGPT subscription (including the free tier) or an OpenAI API key, and a working codex login; usage counts against your normal Codex limits. If you have never touched Codex, /codex:setup can install it and check readiness before you run anything else.
READ THE FULL ANALYSIS
What the three skills actually are. codex-cli-runtime, codex-result-handling, and gpt-5-4-prompting are not things you ask for by name — their own descriptions call them an "internal helper contract" and "internal guidance," and they exist so the slash commands know how to call Codex and present its output back to you. If you came here browsing skills, the product is the eight commands, not these three.
One thing to weigh before enabling the review gate. The optional Stop-hook review gate re-runs Codex on every response and blocks completion until it is satisfied. The README's own warning is that this can create a long-running Claude/Codex loop and drain usage limits quickly, so it is worth turning on only for a session you plan to watch.
ALSO IN THIS PACKAGE
PLUGINS
1 itemPLUGIN
codex
Use Codex from Claude Code to review code or delegate tasks.
AGENT PERSONALITIES
1 itemAGENT PERSONALITY
codex-rescue
Proactively use when Claude Code is stuck, wants a second implementation or diagnosis pass, needs a deeper root-cause investigation, or should hand a substantial coding task to Codex through the shared runtime
PROMPT TEMPLATES
2 itemsPROMPT TEMPLATE
adversarial review
PROMPT TEMPLATE
stop review gate
WHAT'S INSIDE
3 showing · 3 totalReady to use
Nothing else to set up — install it and go.
3 of 3
SETUP
Showing all 3 skills
codex-cli-runtime
When Claude Code hands a stuck coding job over to OpenAI's Codex, this is the rule that keeps Claude out of the way: pass the request on once, and return Codex's answer word for word.
codex-result-handling
Nothing gets fixed until you say so. That is the point of this one: Codex's answer is passed on to you as Codex wrote it, and then the assistant stops.
gpt-5-4-prompting
A request to Codex gets a far better answer when it spells out exactly what finished looks like; this is the house style for writing them that way.
No skill in this package matches that.
HOW TO GET IT
WHOLE PACKAGEInstall the whole package (3 skills)
npx skills add openai/codex-plugin-ccSINGLE SKILLOr install a single skill
npx skills add openai/codex-plugin-cc --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.
PLUGIN MARKETPLACEAdd the marketplace, then install the plugin
/plugin marketplace add openai/codex-plugin-cc/plugin install codex@openai-codexTyped inside the agent's own prompt, not in a terminal. The marketplace is called openai-codex, which is the part after the @.