submit-role

name	submit-role
description	Submit a proof role prompt to a ChatGPT project via CDP and monitor for completion. Verifies Extended Pro (or toggles Deep Research for the literature role), sends, reports the chat URL, and provides recovery guidance for long-running roles.

submit-role

Canonical content: .claude/commands/submit-role.md in this repo. Codex and Claude Code share the same operational steps — edit the Claude command file, not this stub, when changing the procedure.

TL;DR for Codex sessions

Use this when MathPipeProver pauses with waiting_external_agent and the orchestrator needs to fulfill the pending request manually. The skill:

1. Picks the right model mode based on the prompt file — Extended Pro for analytical roles, Deep Research for the literature role (prompts/soft/02_literature_soft.md).
2. Submits via scripts/chatgpt_browser_agent.sh submit ... (preferred) or scripts/chatgpt_browser_agent/cdp_submit.mjs (lower-level).
3. Reports the chat URL.
4. Suggests monitoring tools: /inspect-chat (one-shot), /recover-chat (harvest), /heartbeat <interval> (unattended orchestrator-pace loop).

Read .claude/commands/submit-role.md for the full step-by-step, CLI argument shapes, and error-recovery guidance.