Ops and observability console for my deployed projects. A Node.js/Express server probes each live service server-side, measures real latency and keeps a rolling window, so uptime, p50/p95 latency and sparklines come from actual measurements; metrics not yet instrumented are shown as clearly labelled samples, never fake numbers. Glass design-system UI with a fleet service map and a read-only MCP manifest. Deployed on Fly.io. Code: github.com/abheet19/Helm