Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/openai/codex-plugin-cc/llms.txt

Use this file to discover all available pages before exploring further.

​
Review before shipping

Use /codex:review to get a read-only Codex review of your current uncommitted changes. For a branch review against a base ref, add --base. 
/codex:review

/codex:review --base main
For a review that questions your implementation choices — tradeoffs, failure modes, and alternative approaches — use /codex:adversarial-review instead: 
/codex:adversarial-review --base main challenge whether this was the right caching and retry design
Neither command makes changes to your code. They return Codex’s output verbatim.

​
Hand a problem to Codex

Use /codex:rescue to delegate investigation or a fix request to Codex. Describe what you want Codex to look into. 
/codex:rescue investigate why the build is failing in CI
You can also ask Claude Code to delegate naturally: 
Ask Codex to redesign the database connection to be more resilient.
Claude Code will route the request to the codex:codex-rescue subagent, which forwards it to your local Codex CLI.

​
Start something long-running in the background

When a task is likely to take a while, use --background to let Codex run independently while you continue working. 
/codex:adversarial-review --background
/codex:rescue --background investigate the flaky test
Then check in on the job at any time: 
/codex:status
/codex:result
/codex:status shows all jobs for the current repository with their current state. /codex:result returns the full Codex output once the job completes.

​
Continue a previous Codex task

When you run /codex:rescue after a prior task, the plugin checks whether a resumable thread exists for the current Claude session. If one does, it asks whether to continue it or start fresh. To skip the prompt and always resume the last thread: 
/codex:rescue --resume apply the top fix from the last run
To always start a new thread: 
/codex:rescue --fresh start over with a clean investigation

​
Use a smaller model for quick tasks

Pass --model to select a specific model for one command. Use spark as a shorthand for gpt-5.3-codex-spark. 
/codex:rescue --model spark fix the issue quickly
For more control over reasoning effort, add --effort: 
/codex:rescue --model gpt-5.4-mini --effort medium investigate the flaky integration test
If you omit --model and --effort, Codex uses its own defaults (or whatever you have set in your config.toml).

​
Moving work to Codex directly

After any delegated task or stop-gate run, you can move the session directly into the Codex interface. The Codex session ID appears in the output of /codex:result. 
codex resume <session-id>
This opens the exact Codex run so you can review what it did or continue the work with the full Codex interface.

Build docs developers (and LLMs) love

Get started for freeTalk to us