codeybox-operationslisted

# CodeyBox Operations Use this skill for deployment and operations. Use `AGENTS.md` when changing CodeyBox source, and follow the linked files in `docs/` for full schemas. ## Run a deployment preflight Before changing a production installation, inspect the host and ask about unresolved policy choices. Do not silently omit optional components or choose production defaults on the user's behalf. Ask explicitly: **"Do you want the CodeyBox Admin UI installed?"** Do not infer the answer from an API-only deployment. If yes, confirm its authentication and any exception to the default co-hosted ingress topology. Confirm: - sandbox provider and storage pool; - whether to deploy Admin, its authentication, and any requested ingress exception; - whether provider baselines should be disabled, baked lazily, or prewarmed; - required network profiles and which phases may access the internet; - repository credentials, upstream PR delivery, and auto-merge policy; - persistent state, backups, log retention, and failure notifications. Record non-secret choices in version-controlled configuration. Keep tokens and generated credentials in protected environment files or a secret store. ### Co-host the Admin dashboard by default Use one public CodeyBox origin by default: serve Admin at the origin root and proxy the orchestrator API at `/api/` through the same TLS terminator or tunnel. Keep the API listener loopback-only, and update every caller (including JobTrack) to use the `/api` base UR