Rootly MCP

Rootly's MCP server puts an incident management workspace behind the Model Context Protocol, generating its tools from the platform's REST API so an agent can reach incidents, alerts, services, teams and on-call schedules through ordinary tool calls. An assistant can search incidents by status, severity or service, read the timeline and action items attached to one, look up which team owns a failing service, and check who is on call for a given rotation. Because the server is generated from the API surface, it also supports the write side of response work: opening an incident, updating its status or severity, adding timeline entries, and assigning roles without leaving the conversation where the debugging is already happening. Teams commonly pair it with a coding agent so the session reading stack traces is the same one filing and updating the incident record. The audience is on-call engineers, SRE teams and incident commanders running response in Rootly, plus platform teams building automation on top of it. The server is open source and authenticates with a Rootly API token, which bounds what an agent can read or change to that token's permissions. It runs locally over stdio or as a hosted remote endpoint, so it fits Claude Code, Claude Desktop, Cursor and other MCP clients.