Observability that runs anywhere you can run a binary.
Send your traces, logs, and metrics over OpenTelemetry. Get a fast UI, a chat investigator, and alerts — without standing up a stack.
| Service | Health | p95 | Errors | Throughput |
|---|---|---|---|---|
| checkout | unhealthy | 412 ms ↑38% | 2.4% | 1.2k rps |
| cart | degraded | 187 ms | 0.6% | 2.1k rps |
| catalog | healthy | 42 ms | 0.0% | 3.8k rps |
| payment | healthy | 94 ms | 0.1% | 980 rps |
| auth | healthy | 28 ms | 0.0% | 5.4k rps |
Everything you'd build yourself, in one process.
No collector to deploy. No database to operate. No SaaS to budget for. Just a binary that does the work.
Speaks OpenTelemetry, day one.
Point any OTel SDK or collector at it. Traces, logs, metrics — all on the standard wire format.
Sub-second over months of data.
Designed to keep up with high-volume telemetry without an external database, cluster, or sidecar.
Just ask: "why is checkout slow?"
A chat investigator built into the binary. Connects to Claude Code over MCP for hands-on triage.
Self-hosted by design.
Your data stays on your network. No per-host fees, no telemetry tax, no agents shipping data offsite.
Just ask.
A chat investigator is built into the binary. Hook it to Claude Code over MCP and triage from your editor — or use it from the browser.
Three commands. You're observable.
Install
Grab the binary or run the container. One file, no dependencies, <30MB.
Point your apps
Set OTEL_EXPORTER_OTLP_ENDPOINT and you're streaming.
Open the UI
Service map, traces, logs, alerts — already there, already wired up.
Try it in 30 seconds.
One binary. No signup. Your data stays on your machine.