Self-hosted cluster monitoring
Linux, macOS and Windows hosts reporting to a dashboard you run. See pooled capacity across the whole cluster, then drill into any host. No inbound firewall rules, no SaaS, no per-host pricing.
Your whole fleet as one machine: pooled cores, RAM and disk in a single line.
Then per host, with thresholds flagging trouble before you go looking for it.
Cluster totals
Pooled CPU, RAM and disk across every host — and CPU is core-weighted, so a busy 32-core box isn’t diluted by an idle 2-core one.
Push, not pull
Hosts behind NAT or on Tailscale report in. No collector reachability to arrange.
Load average, not just %
Utilisation pins at 100% and goes quiet. Load keeps climbing — per host, per core.
Port activity
Per-port connection counts with history, so you can see which service is busy.
Your data, your box
SQLite on hardware you control. Plain JSON API to script against.
Per-host detail: meters, thresholds, load and the ports taking traffic.
1. Run the agent → 2. Paste the ingest token → 3. The host appears
Unlock the full app: one-time purchase via Stripe, then download the installer. Your GitHub account is linked to your license.
1. Sign in ✓ → 2. Purchase → 3. Download
Dashboard
Send one verified Docker release to every registered Mothership agent. Each host pulls over HTTPS, verifies the SHA-256 checksum, then activates its locally configured container deployment hook.
Create a profile to get an ingest token for a new host, then add it to
that host's agent.conf as MOTHERSHIP_INGEST_URL. Point the URL at
this Mothership Server's own tailnet or LAN address on port 8097 when possible. Remote clients
without that route can use this public HTTPS hostname at /api/ingest/<token>;
only that bearer-token ingest path is exposed, while dashboard reads remain signed-in and
entitlement-gated. (The preferred tailnet address is visible on this Mac's own card above.)
Widgets
Dashboard widgets
API
Mothership API access