PagerDuty MCP

The PagerDuty MCP server is the vendor's own Model Context Protocol bridge to its incident response platform. It exposes the objects an on-call engineer works with every day as agent-callable tools: incidents, alerts, services, escalation policies, on-call schedules, teams, users and incident notes. Authentication uses a PagerDuty API token, and the token's scope decides what the agent can read and what it can change, so a read-only key produces a safe triage assistant while a full key allows write actions. With it connected, an assistant can answer questions like which incidents are open for a given service, who is on call for a rotation, and what the timeline of log entries on an incident looks like, then take action: acknowledge, reassign, escalate, add a status note or resolve. Because responses come back as structured records rather than page scrapes, the agent can correlate incidents across services, summarize an outage for a status update, or draft a postmortem from the incident log without a human copying data out of the console. It is aimed at SRE, DevOps and platform teams who already run PagerDuty and want triage from an IDE, terminal or chat client. The server is published as open source, so teams can audit exactly which API calls each tool makes before granting it a production token.