HomeLab Monitor
Inspect homelab hosts, Docker, GPU, and services from your AI agent via a read-only MCP server bundled with HomeLab Monitor.
Overview
io.github.SikamikanikoBG/homelab-monitor is an MCP server for the Operate phase that provides read-only homelab monitoring for hosts, Docker, GPU, and services via HomeLab Monitor.
What is this MCP server?
- 12 read-only MCP tools: hosts, containers, GPU, memory, services, alerts, events, history
- Ships inside the same Docker image as the HomeLab Monitor dashboard (v0.14.2)
- Streamable HTTP on port 9810 or stdio via docker run with HOMELAB_MONITOR_URL
- MIT licensed, self-hosted—no cloud dependency for homelab observability
- 12 read-only MCP tools (list_hosts, get_containers, get_gpu, get_alerts, and others)
- Bundled MCP from HomeLab Monitor v0.14.0+; registry version 0.14.2
- MIT license; streamable HTTP on hub :9810/mcp
Community signal: 64 GitHub stars.
What problem does it solve?
Your agent cannot see Docker, GPU, or host health on your homelab without you pasting command output from every machine.
Who is it for?
Solo builders with a self-hosted HomeLab Monitor hub who want AI-guided triage on local AI rigs and Docker fleets.
Skip if: Teams on pure cloud PaaS with no homelab monitor URL, or anyone needing write/actions on containers from MCP.
What do I get? / Deliverables
Your assistant queries 12 read-only MCP tools against HomeLab Monitor for a unified snapshot of containers, services, GPU, and alerts.
- Agent-accessible host, container, GPU, and service snapshots
- Alert and event context from your homelab fleet
- Disk scan and history queries without manual SSH
Recommended MCP Servers
Journey fit
Homelab fleet visibility is an operate concern once you run self-hosted AI and services day to day. The built-in MCP exposes 12 read-only monitoring tools over HTTP or stdio so agents can snapshot infra without SSH sprawl.
How it compares
Read-only homelab observability MCP, not a cloud provisioning or deploy skill.
Common Questions / FAQ
Who is HomeLab Monitor MCP for?
Homelab and indie AI operators who run HomeLab Monitor and want agents to read host, Docker, GPU, and service status safely.
When should I use HomeLab Monitor MCP?
During operate when you monitor self-hosted services and need quick fleet snapshots inside Claude or Cursor.
How do I add HomeLab Monitor MCP to my agent?
Run HomeLab Monitor, then add HTTP at http://YOUR-HUB:9810/mcp or stdio via docker with HOMELAB_MONITOR_URL pointing at your hub.