What problem does it solve? Running a local-first AI runtime day to day involves many moving parts: starting the Ollama-backed API server, draining it safely, managing PID files and logs, taking backups, and applying signed updates. This Skill provides a copy-pasteable operational runbook so you never have to guess which subcommand, port, or state path applies. ## Core Features & Use Cases - Lifecycle Management: Start, stop, restart, and gracefully drain the Sonder API server (port 11435), the Ollama daemon (11434), and the mobile-control launcher (11436) via python -m sonder_runtime subcommands, Windows/POSIX launchers, or the headless supervisor. - Backup, Restore, and Updates: Create, verify, prune, and apply backups; install signature-verified update bundles with rollback; or run guarded fast-forward updates on source checkouts. - State and Deployment Knowledge: Know exactly where databases, logs, and PID files land per OS, and deploy as a loopback-only systemd service on Linux. - Use Case: You need to restart a production Sonder server after a schema migration. The Skill tells you to run migrate --adopt-epoch2, then serve, and how to verify with a smoke test against /v1/chat/completions. ## Quick Start Ask the assistant to start the Sonder API server and verify it is responding with a smoke test.