pagerduty-mcp-server

Adoption verdict: if you already use PagerDuty and want an MCP-compatible AI agent to directly read incident, service, schedule, and on-call data—and possibly perform acknowledge or resolve actions—this project is worth tracking. But the evidence is extremely limited: almost everything comes from the GitHub repo description, which is not enough to prove maturity, reliability, or real-world adoption.

In practice, this looks more like a PagerDuty-to-agent connector layer than a full AIOps platform, alert aggregator, or standalone support bot. A better analogy is an MCP wrapper around PagerDuty resources and actions, so clients like Claude Desktop or other MCP-capable agents can call incidents, services, schedules, on-calls, and selected write actions.

On setup and cost, the supported claims are modest: you likely need a PagerDuty environment, OAuth or an API key, and an MCP client configuration. The repo snippet supports what objects and actions are exposed and the auth methods, but it does not support stronger claims about deployment difficulty, permission boundaries, production readiness, or API pricing.