# Pitot MCP server

> Allows AI agents to investigate OpenTelemetry traces, metrics, and logs using typed query tools and dashboards.

- Repository: https://github.com/labstack/fanout
- Page: https://hostedmcp.org/servers/labstack/fanout
- Official: yes
- Transport: remote
- Stars: 120
- License: Apache-2.0
- Latest version: 2026.11.3

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `OTEL_EXPORTER_OTLP_ENDPOINT` | no | no |  |
| `OTEL_EXPORTER_OTLP_HEADERS` | no | no |  |
| `OTEL_EXPORTER_OTLP_PROTOCOL` | no | no |  |
| `PITOT_AUTH_CODE_SECRET` | no | no |  |

## FAQ

### How can I connect an MCP client to Pitot?

Set PITOT_PUBLIC_URL to your stable HTTPS origin and point the client at the /mcp endpoint. The client uses OAuth for discovery and registration.

### Does Pitot require external storage services?

No, Pitot keeps traces, logs, and metrics in a single binary and one data directory with no external storage services required.

### Can I use AI to query my telemetry?

Yes, you can enable chat by selecting an AI provider like Anthropic or OpenAI and providing the necessary API key.

### What protocols does Pitot support for ingestion?

Pitot supports OTLP ingestion over both gRPC and HTTP.