codex-cli-runtime

Codex Runtime

Use this skill only inside the codex-router:codex-rescue subagent.

Primary helper:

• node "${CLAUDE_PLUGIN_ROOT}/scripts/codex-companion.mjs" task --watch "<raw arguments>"

Execution rules:

• The rescue subagent is a forwarder, not an orchestrator. Its only job is to invoke task --watch once and return that stdout unchanged when the watched job finishes.
• Prefer the helper over hand-rolled git, direct Codex CLI strings, or any other Bash activity.
• Do not call setup, review, adversarial-review, status, result, or cancel from codex-router:codex-rescue.
• Use task for every rescue request, including diagnosis, planning, research, and explicit fix requests.
• You may use the gpt-5-4-prompting skill to rewrite the user's request into a tighter Codex prompt before the single task call.
• That prompt drafting is the only Claude-side work allowed. Do not inspect the repo, solve the task yourself, or add independent analysis outside the forwarded prompt text.
• Leave --effort unset unless the user explicitly requests a specific effort.
• Leave model unset by default. Add --model only when the user explicitly asks for one.
• Preserve model selectors and aliases such as spark; the runtime resolves them against the live catalog.
• Default to a write-capable Codex run by adding --write unless the user explicitly asks for read-only behavior or only wants review, diagnosis, or research without edits.

Command selection:

• Use exactly one task invocation per rescue handoff.
• If the forwarded request includes --background or --wait, treat that as Claude-side execution control only. Strip it before calling task, and do not treat it as part of the natural-language task text.
• Always call internal task --watch without --background. The outer command backgrounds the rescue subagent when requested. --watch launches a detached tracked worker, captures its exact job id, and reconciles only that authorized job.
• The watcher may expire, but the active worker must continue. Never translate watcher or Bash expiration into job cancellation or failure.
• If the forwarded request includes --model, pass the selector through unchanged to task; the runtime resolves exact slugs and live aliases.
• If the forwarded request includes --effort, pass it through to task.
• If the forwarded request includes -c/--config, --enable, or --disable, pass those controls through to task.
• If the forwarded request includes --resume, strip that token from the task text and add --resume-last.
• If the forwarded request includes --fresh, strip that token from the task text and do not add --resume-last.
• --resume: always use task --resume-last, even if the request text is ambiguous.
• --fresh: always use a fresh task run, even if the request sounds like a follow-up.
• --effort: pass through the requested level. Do not maintain a fixed list; Codex effort names evolve. When a model is explicit, the runtime validates the level against that model's live codex debug models entry.
• task --resume-last: internal helper for "keep going", "resume", "apply the top fix", or "dig deeper" after a previous rescue run.

Safety rules:

• Default to write-capable Codex work in codex-router:codex-rescue unless the user explicitly asks for read-only behavior.
• Preserve the user's task text as-is apart from stripping routing flags.
• Do not inspect the repository, read files, grep, monitor progress, poll status, fetch results, cancel jobs, summarize output, or do any follow-up work of your own.
• Return the stdout of the task --watch command exactly as-is when it completes.
• If Bash expires after reporting Codex rescue started as <job-id>, return that exact job id and /codex-router:result <job-id>, and state that the active job was not cancelled.
• If the Bash call fails before a job id is reported or Codex cannot be invoked, return nothing.