PagerDuty MCP server: setup, auth, production

What is the PagerDuty MCP server?

A PagerDuty MCP server implements the Model Context Protocol and exposes PagerDuty's capabilities as typed tools an agent can discover and call in the middle of a task. Servers here are community-maintained, which makes provenance and pinning part of the setup rather than an afterthought. What this setup has in common with every other integration in the Observability category is covered under Observability on the MCP index.

A real PagerDuty workflow

The agent summarizes active incidents, tells you who’s on call for the affected service, and drafts an incident timeline from the event stream — situational awareness assembled while you’re still reading the first alert.

What an agent can do with PagerDuty

• Have an agent summarize active incidents
• Pull who’s on call for a service during an investigation
• Draft an incident timeline from events

Read-only is the right default: let the agent observe and reason first, then grant specific write actions deliberately — each behind audit logging and, for anything high-impact, human approval.

Setting up the PagerDuty MCP server

Use a PagerDuty API token scoped to read access. Resist the urge to grant write: an agent should inform incident response, not act inside it.

Auth: A PagerDuty API token scoped to read access. Credentials belong in the environment or a secret store — never in the repo or the agent's sandbox. Wiring a server into Claude Code or Cursor is the same for every tool; the step-by-step lives in the MCP connection guide.

The risk to watch with PagerDuty

An agent that can acknowledge or resolve incidents can hide a live problem. Read-only, always — the on-call human owns acknowledge and resolve.

Scoping a PagerDuty server down

Read access to incidents, services and schedules is the useful part and is safe. The write that matters here is anything that changes incident state or on-call assignment, and it should require a human without exception: an agent that acknowledges an incident has silenced a page someone needed to receive, and an agent that overrides a schedule has altered who gets woken up. Scope the API key to read-only and to the specific services in scope. During an incident the agent's value is assembling context quickly, which read access fully supports.

More broadly, a production-grade PagerDuty integration needs read scoping and never auto-acknowledging or resolving incidents. The MCP security checklist applies to every integration — we build servers with these controls baked in.