deploylisted

# Deploy Run a project's **production deploy** by executing its documented, ordered playbook — confirm-gated and complete. The whole point of this skill is that **nothing important gets skipped and nothing gets improvised**: the playbook is authoritative; if reality diverges from it, you stop and ask rather than guess. > Deploy touches production. Expect `sudo` / `systemctl` steps to raise approval prompts — > that is correct and desired here (a human is present). Follow § Shell hygiene in > `procedures.md`: one command per Bash call, `git -C <path>` / `sudo -u <user> …`, never > `cd <path> && …`. --- ## Step 0 — Identify the project and load its playbook 1. Determine the target project (from the command argument, else ask). `Read config/projects/<dir>/context.md`. 2. Locate the **authoritative, ordered deploy playbook**, in this order: - the `## Deployment` section of `context.md` (structured per the playbook contract in `procedures.md` → *Deployment — the `/deploy` playbook contract*), and - the repo document it points to for exact commands (e.g. polimarket-analyser → `README.md` § *Deploy a merged `dev`* + *Gotchas*). Read it in full. 3. **If the playbook is missing or an unauthored stub, do not deploy yet:** - **No deploy section at all** → STOP. Tell the human the project has no deploy playbook and do not improvise one. Point them at the contract in `procedures.md`. - **A stub / unauthored placeholder** (marked 🚧, or its parts are `TOD