A local macOS command center that watches an entire project fleet from one pane of glass. Live status lights, KPI sparklines, git freshness, uptime, and launchd schedule checks all stream in from a 24/7 Node daemon, rendered behind a NASA-ops interface.
The problem
Run a dozen projects at once and their health scatters across terminals, cron logs, and git checkouts. You find out something broke long after it actually did. There's no single place to glance at and know which projects are fresh, running, and on schedule.
What I built
Built and running locally. It monitors 12 projects in one Fleet Grid with live status lights and KPI sparklines. The audit agents have already surfaced 5 real findings, proven on the troy project, with no write access to any monitored repo.
How it works
A 24/7 Node daemon runs real probes against each project: git freshness, uptime, launchd schedule state, KPIs. It writes the results to a node:sqlite telemetry store. The Next.js 15 / React 19 front end, written in TypeScript, renders them as a read-only Fleet Grid and pushes updates over SSE every 30 seconds. Boot starts with an env safety gate, a self-test that fails closed if it detects a secret leak before the daemon comes up. On top of that sits the M3 hole-finder. It runs suggest-only audit agents that spawn the trusted Claude CLI in read-only mode and hand back findings instead of touching code.
Highlights
- Read-only Fleet Grid with live status lights and KPI sparklines
- Git freshness, uptime, and launchd schedule probes every 30s
- 24/7 Node daemon writes telemetry to a node:sqlite store, streamed over SSE
- Env safety gate: boot self-test fails closed on secret leaks
- Suggest-only audit agents spawn a read-only Claude CLI and return findings
- Monitors 12 projects; audit agents surfaced 5 real findings on troy
Let's build yours.
Tell me what you're trying to ship — you'll get a scoped plan and a straight answer.