I have a dedicated Mac Mini that must run OpenClaw around the clock as a fully autonomous AI agent focused on workflow automation for our internal business processes. I need more than a quick install—I’m after rock-solid configuration, reliable persistence on reboot, and clear observability so the box can sit in a server rack and quietly do its job without surprises. Here is what success looks like to me: • OpenClaw installed with all required dependencies, tuned for macOS and the machine’s hardware resources. • Launch-at-boot service configured, with graceful shutdown/restart routines. • Logging, alerting, and basic health metrics surfaced in a way that can plug into the monitoring stack we already use (Prometheus/Grafana or an equivalent you recommend). • A starter workflow that proves the end-to-end pipeline: the agent picks up a trigger, executes a defined automation task, and records the result. I’ll provide remote access to the Mac Mini plus any API keys or internal endpoints the agent must call. You bring deep familiarity with AI agents, Python dependency management, and macOS background services (launchd, Homebrew, etc.). Clear documentation—commands, config files, and a short runbook—rounds out the hand-off. If this reads like your wheelhouse, tell me how you’d approach hardening OpenClaw for 24/7 use and which monitoring approach you favour.