incident.io MCP

The incident.io MCP server is a vendor-maintained remote endpoint that lets MCP-compatible assistants read and act on an incident management workspace over the Model Context Protocol. Once connected with an API key or through the hosted OAuth flow, an agent can list open and historical incidents, pull the full timeline of updates, actions and follow-ups attached to one, look up the alerts and alert routes that triggered it, and check who is on call across escalation policies and schedules. Access is scoped by the credential you connect with, so the server inherits the same permissions model as the underlying API rather than granting blanket access. Read-heavy workflows are the common starting point: asking an assistant to summarize what happened during an outage, group recurring alert sources, or draft a postmortem from the recorded timeline instead of copying it out of the web app by hand. Write capabilities extend that to creating and updating incidents and follow-up items from the same conversation. It is aimed at on-call engineers, SRE and platform teams, and incident commanders who already run response in incident.io and want their coding agent or chat assistant to answer questions about live and past incidents without a browser detour. Because it speaks standard MCP, it drops into Claude Code, Claude Desktop, Cursor and other clients that support remote servers.