Homelab Health Dashboard

A single n8n workflow that SSHes into your server, collects 30+ system and Docker metrics, compares them against 7 days of historical data, and has GPT-4o-mini analyze everything with structured, actionable findings.

### The Problem

If you run a homelab, you have done this: SSH in, run htop, check docker ps, glance at disk usage, decide everything is fine, close the terminal. Maybe you do it daily. Maybe you skip weekends. And then you wake up to a dead service because the disk filled up or a container entered a restart loop at 3 AM.

Full monitoring stacks (Prometheus + Grafana + Node Exporter + Alertmanager) solve this properly, but they add 3-4 containers and real configuration overhead. For a single server running a handful of Docker services, that is often more monitoring infrastructure than actual infrastructure.

### The Solution

This workflow SSHes into your server, collects real CPU utilization %, all mounted filesystems, network I/O, top processes, Docker ecosystem health (system df, dangling images, reclaimable space), and 20+ other metrics. It compares everything against 7 days of history stored in Google Sheets, then sends it to GPT-4o-mini.

The AI returns structured JSON with severity-tagged findings -- each with a title, detail explanation, and the exact CLI command to fix it. The digest lands in Discord as a 4-embed dashboard you can scan in seconds: status header with inline metric fields, severity-tagged findings with code blocks, Docker ecosystem stats + trends, and a footer with timestamp and API cost.

A parallel trigger runs every 5 minutes checking critical thresholds. Disk over 90%, memory over 95%, any container down -- instant Discord alert with the command to fix it.

A first-time setup trigger creates a formatted Google Sheet with dark headers, frozen rows, and conditional formatting gradients for 6 key metrics. One click, no manual sheet creation.