AGENTS.md@commander/cowork-plugin-codex git:20261013.fbe187dA

Immutable. This exact content never changes and is served at /api/v1/blob/09bd2cb8e672d55f.

# AGENTS.md — CC Commander (Codex / ChatGPT Work mirror)

> Generated by `npm run build:codex` from `commander/contract.json` v8.0.0. Do not edit by hand.

## What this directory is

The Codex-format mirror of the CC Commander plugin: 84 skills (`skills/*/SKILL.md`, translated from the Claude Code tree), 22 specialist agent personas (`agents/*.toml`), the Codex-supported hook subset (`hooks.json`), and 2 credential-free MCP servers (`.mcp.json`).

## How to use it

- Start with the `ccc` skill (`$ccc`) — the guided hub for the /ccc-* workflows.
- Skills load from `skills/`; invoke explicitly with `$<skill-name>` or let the task match implicitly.
- Agent personas in `agents/*.toml` define specialist voices (architect, reviewer, security-auditor, …). Adopt the matching persona when a task fits one.
- Skill/hook commands reference `${CLAUDE_PLUGIN_ROOT}` — Codex exposes it as a documented compatibility alias for its native `PLUGIN_ROOT`, so it resolves correctly here unmodified.

## Operating rules for agents working here

- This tree is GENERATED. Never edit it directly — change `commander/cowork-plugin/` and run `npm run build:codex`.
- Plan before building; verify before declaring done. Prefer small, reviewed diffs.
- Never commit secrets. Use env vars for anything sensitive.
- Run the repo gates before pushing: `npm test`, `node scripts/check-product-contract.js --check`, `node scripts/check-compat.js`.

## Links

- Homepage: https://commanderplugin.com
- Repo: https://github.com/KevinZai/commander
- Compat matrix (Claude Cowork / Claude Code CLI / Codex CLI / ChatGPT Work): `docs/compat/chatgpt-work.md`