claude-to-codex-coordinatorlisted

Coordinate Claude Code-to-Codex delegation through OpenAI's official codex-plugin-cc runtime. Use when Claude should ask Codex to review, investigate, implement, resume, transfer a Claude session, run in the background, report status or results, or cancel work while preserving scope, permissions, model intent, and evidence.
wildcard/codex-to-claude-coordinator · ★ 0 · AI & Automation · score 72

Install: claude install-skill wildcard/codex-to-claude-coordinator

# Claude to Codex Coordinator Use `coordination-core` for classification and lifecycle envelopes, then map the result to OpenAI's official `codex@openai-codex` Claude Code plugin. Keep the portable contract independent of the plugin's commands and state strings. Read [references/openai-codex-plugin-cc.md](references/openai-codex-plugin-cc.md) before installing, upgrading, transferring a transcript, enabling the stop review gate, or relying on a specific command or runtime behavior. ## Probe Before Delegating When loaded as this Claude Code plugin, run: ```sh python3 "${CLAUDE_PLUGIN_ROOT}/skills/claude-to-codex-coordinator/scripts/probe.py" ``` In another skills-compatible harness, resolve `scripts/probe.py` from the installed directory that contains this `SKILL.md`; never resolve it from the active project's working directory. The probe reports only sanitized availability facts. It omits install paths and raw authentication output. Stop and resolve any false prerequisite before delegating. Do not copy raw `/codex:setup` output into a repository or handoff. It can contain account labels. Do not enable the optional stop review gate during setup unless the user explicitly requests it and will monitor the resulting Claude/Codex loop. ## Resolve the Assignment 1. Apply the portable six-axis classifier. 2. Record whether the task is review-only or write-capable. 3. State the repository scope, definition of done, required tests, evidence, and external-action boundary.