This is plumbing, not a skill you'll call directly. It lives inside the codex-rescue subagent and does one thing: forward your request to the Codex companion runtime via a single task invocation, then return the output unchanged. The rescue agent isn't allowed to inspect your repo, orchestrate steps, or summarize results. It just rewrites your prompt if needed, maps flags like --resume or --model spark, defaults to write mode unless you ask otherwise, and hands everything to Codex. If you're using Claude Code's Codex integration, this is the internal contract that makes the handoff work. You won't invoke it yourself.
npx skills add https://github.com/openai/codex-plugin-cc --skill codex-cli-runtime