PagerDuty Agent

What you can do

A few things this template does well out of the box.

• Triage open incidents — group by service, surface owners, see the latest comments
• "Who's on call for the `payments` service right now? Who's next on the rotation?"
• Pull incident history for a service — frequency, MTTR trend, top recurring causes
• Acknowledge or resolve an incident from chat (with confirmation)
• Generate a stand-up summary: incidents from the last 24 hours, by service, with current status

How it works

Three steps to go from template to a live chat.

MCP servers used

The endpoints this template connects to by default. You can swap any of them in Agent Studio.

Getting your access token

A quick walkthrough for the credential this template needs.

1. 1**Generate an API token**: PagerDuty → **User Icon → My Profile → User Settings → API Access Keys** → create a new key. (For org-wide access, an admin can create one under **Configuration → API Access Keys** instead.)
2. 2Paste the token into Agent Studio's **Access token** field. The default URL `https://mcp.pagerduty.com/mcp` is correct for all PagerDuty regions.
3. 3**OAuth alternative**: if your org uses SSO, the same URL supports OAuth 2.0 — you'll be prompted to authorise on first connection.
4. 4**Permissions**: the token inherits the user's role. For triage-only workflows, create the key under a Restricted Access user. For incident-management workflows (ack/resolve), use a Responder or higher role.
5. 5Send *"List all open incidents right now"* to confirm.

Try these prompts

Copy one into the studio to see the agent in action.

• List all open incidents right now, grouped by service. Show me severity and the latest comment.

• Who is on call for the `checkout` service right now, and who is the next person on the rotation?

• Pull every incident for the `payments` service in the last 30 days. What is the MTTR? What are the top 3 recurring causes?

• Generate a stand-up summary: every incident triggered between 10pm and 8am, grouped by service.

• For incident PD-12345: who acked it, what was the timeline, and is there a postmortem?

System prompt

The default instructions the model starts with. Edit it any time inside Agent Studio.

You are a senior incident response engineer connected to PagerDuty via the official hosted MCP server. You help on-call engineers triage incidents and run the on-call workflow.

Use the available tools to:
- List and search incidents (open, acknowledged, resolved) with filters by service, severity, time range
- Read incident details: timeline, comments, assignee, escalation state, linked runbooks
- Query on-call schedules: who's on call now, who's next, who covered which window
- Inspect services, escalation policies and integrations
- Acknowledge, resolve, snooze or reassign incidents (with explicit user confirmation)

Operating principles:
- Lead with the answer ("3 open incidents on payments, all triggered in the last 20 minutes"), then back it up with the details
- For triage views, group incidents by service and surface common patterns — not a flat list
- For "who's on call" questions, always include the next person in the rotation and when the handover is
- Never acknowledge / resolve / snooze an incident without the user explicitly asking for it
- For postmortem questions, point at the linked postmortem doc instead of inventing a summary