FleetView

Open-source monitoring for a fleet of Starlink dishes. Live telemetry, obstruction maps and alerting for every site — maritime, mining, remote camps, ISP relays.

No dish? No problem.

The hard part of building anything for Starlink is that you need a dish — ideally forty of them, in bad weather, on the other side of the planet.

So FleetView ships its own fleet. One command starts 40 simulated dishes that speak the real Starlink local gRPC protocol, each with believable telemetry, plus fault injection you can trigger from a terminal:

npm run setup   # once
npm run dev     # 40 dishes + agent + cloud + dashboard

Open http://localhost:5273. That's the screenshot above, running on your machine in about ten seconds — no hardware, no accounts, no API keys, no Docker.

Point it at a real dish when you have one. It's the same agent either way.

Break things on purpose

Every simulated dish has a control plane. Knock a vessel offline and watch the map turn red, the alert fire, and the site drop into "needs attention":

curl -X POST http://127.0.0.1:8799/scenario \
  -H 'content-type: application/json' \
  -d '{"id":"mv-pacific-voyager","mode":"obstructed"}'

Modes: normal · offline · obstructed · high_latency · degraded · thermal Reset everything with curl -X POST http://127.0.0.1:8799/scenario/reset.

This is the bit that's genuinely hard to get any other way. Reproducing a thermal shutdown, a mid-poll reboot, or a tree growing into the sky view is otherwise a matter of waiting months and getting lucky.

What you get

• Live, not lagged — telemetry lands in seconds, over a WebSocket.
• Any dish — reads the local API directly, so no reseller/account-manager gating.
• Store-and-forward — the agent buffers to local SQLite; a site that loses its uplink backfills when it returns instead of leaving a hole in your charts.
• **Mobile-f