Chronosphere MCP

Chronosphere's MCP server exposes its observability platform as MCP tools so an assistant can run an investigation directly against production telemetry. Agents can query metrics, search logs, pull distributed traces, and list change events, the deploy and configuration changes Chronosphere records so incidents can be correlated with what shipped. Platform objects are readable too: dashboards, monitors, and shaping rules including the drop rules that govern which telemetry is retained. The server is strictly read-only. It supports no mutation actions, so an agent cannot update a dashboard, write metrics or logs, or delete a monitor. That boundary is the selling point for on-call teams: an engineer can hand an agent an alert and let it correlate a latency spike with a recent change event and the traces behind it, with no path for the agent to alter alerting configuration while it works. A companion documentation MCP server makes Chronosphere's own docs searchable in the same session, so an agent can look up query syntax or a shaping concept without the engineer switching tabs. Built for SRE, observability and platform engineering teams already running Chronosphere who want incident triage and dashboard archaeology happening inside the coding assistant they use the rest of the day.