// the find
rcourtman/Pulse
Real-time monitoring for Proxmox, Docker, and Kubernetes with AI-powered insights, smart alerts, and a beautiful unified dashboard
Pulse is a self-hosted monitoring dashboard for Proxmox (VE, PBS, PMG), Docker, and Kubernetes that runs as a single container or LXC on your homelab. It adds an AI chat layer (BYOK) on top of standard metrics polling, with a React frontend and a Go backend. Aimed at homelabs and small MSPs who want something between Grafana and a full enterprise stack.
The Proxmox coverage is unusually thorough — VE, PBS, and PMG all in one place is rare; most alternatives handle only VE. The Patrol feature (scheduled background health checks via LLM) is a legitimately useful idea: it catches things like silently failed backups or ZFS pools creeping toward capacity without requiring you to write alert rules for every edge case. The architecture is sensible — Go backend, SQLite for persistence, agents deployed separately so the main server doesn't need host access. Auto-discovery of Proxmox nodes on the network removes the most tedious part of initial setup.
The freemium split is awkward: auto-fix, Kubernetes AI analysis, and audit webhooks are Pro-only, so the free tier's AI story is just a chat box you have to fund yourself with an API key. The Go backend is a single binary with no published API stability guarantees, which matters if you want to script against it long-term — the API docs exist but the versioning story is unclear. Agent updates use a checksum-verified curl-pipe-to-bash pattern (the very install.sh one-liner in the README), which is the thing the security docs exist to justify but won't satisfy anyone with a stricter threat model. Test coverage looks thin on the backend side; the visible test files are almost all frontend TypeScript, so the Go data-collection paths are harder to trust without reading the source.