SigNoz MCP

SigNoz maintains an MCP server for its open source observability stack, giving an assistant access to all three telemetry signals through one connection. An agent can list the services sending data, pull metrics over a time window, search traces by service, operation, duration or error status, and filter logs by attribute. Because SigNoz is OpenTelemetry native, the filters an agent uses map to the same attribute names visible in the interface. The server also reaches the account layer around the raw telemetry: dashboards and their panels, saved views, and alert rules together with which ones are firing. An agent can therefore answer what is alerting and which service sits behind it, then drill into the traces and logs for the same time window without a context switch or a second tool. It is built for SRE and backend teams running SigNoz, whether self-hosted or on the managed cloud, and connects with an API key. Practical uses include incident triage, latency investigations where the model correlates a metric spike with slow spans and the errors logged alongside them, and quick post-deploy checks on whether error rates or p99 latency moved.