How to Connect PagerDuty to Claude (and What It Can't Do)
PagerDuty runs an official MCP server, so Claude can genuinely read and write your incident data — pull on-call schedules, list open incidents, even create and update incidents — all in plain English. The catch is the one that defines every Claude connection: it only works inside a chat you’ve opened. An incident firing at 3am triggers nothing. Claude has no event listener, no schedule, no way to know a P1 just paged until you sit down and ask.
Below: what the official server exposes, the setup, why “no triggers” is the wrong shape for incident response, and how to get the fan-out — Slack, Jira, stakeholder email, postmortem scheduling — to happen the moment an incident fires.
What the official PagerDuty MCP server exposes
The PagerDuty MCP server is published by PagerDuty itself — no community wrapper to audit — and it’s genuinely read/write, which puts it ahead of the many connectors that only read. It comes in two forms: a PagerDuty-hosted endpoint at mcp.pagerduty.com/mcp, and an open-source self-hosted build you can run in your own environment.
Connected, Claude can:
- Query on-call. “Who’s on call for the payments service right now, and when does their shift end?” — useful before a risky deploy.
- Read incidents. List what’s triggered, acknowledged, and resolved over any window; pull a full incident timeline for a postmortem.
- Manage incidents and policies. Create and update incidents, and work with escalation policies and services.
PagerDuty also shipped a Claude Code plugin in Anthropic’s marketplace that scores uncommitted code against historical incident data — a nice pre-deploy safety check, but a developer-desk tool, not an on-call responder.
Connecting it takes about five minutes
PagerDuty isn’t a one-click app in Claude’s connector directory yet, so you add it as a custom connector:
- In Claude, open Settings → Connectors and choose Add custom connector.
- Paste
https://mcp.pagerduty.com/mcp(or point at your self-hosted URL). - Authorize against your PagerDuty account when Claude bounces you to sign in.
- Ask “list open incidents on the checkout service” to confirm it’s live.
Custom connectors are gated to paid Claude plans, so a free-tier account can’t add the server.
Why “no triggers” is the wrong shape for incident response
Incident response is defined by one thing: something happens, and a chain of actions has to fire immediately, whether or not a human is watching. That is exactly the shape Claude connectors don’t have.
- It reacts to you, not to the incident. A P1 pages at 3am. Claude can create an incident, sure — if you’re awake, in a chat, and typing. It cannot notice the page and start working. There’s no event trigger and no schedule; the connector is a tool you operate, not an agent that runs.
- The fan-out is manual. When a real incident fires, someone posts to the war-room Slack channel, opens the Jira ticket, emails stakeholders, and books the postmortem. Claude can help you draft each of those — one prompt at a time, while you’re the one holding it all together during the fire.
- Nothing persists between chats. Close the tab and the context is gone. The connector isn’t sitting on your PagerDuty account watching for the next page.
So Claude is a sharp desk-side analyst for “what’s the on-call situation and what happened in that incident?” — and structurally can’t be the thing that runs the response while you focus on the fix.
Incident fan-out that fires on its own: Carly
The valuable half of incident response is the reflex: the instant an incident triggers, the right people get told and the right records get created — no one manually fanning it out mid-crisis. Carly — an AI executive assistant that runs on triggers, in the cloud — handles that half:
- A PagerDuty incident triggers on the payments service → Carly posts a formatted alert to the
#incidentsSlack channel, opens a Jira ticket pre-filled with the incident details, and emails the on-call lead and the eng manager — all before you’ve finished reading the page. - The incident resolves → Carly emails the stakeholder list a plain-English summary (what broke, duration, impact), and schedules the postmortem on the responders’ calendars for the next business morning.
- You describe it in plain English — “when a P1 fires in PagerDuty, alert Slack, open the Jira ticket, email stakeholders, and book the postmortem” — and Carly interviews you, then builds the workflow with you. Nothing to host, no connector config.
AI agents start at $35/month, and workflow steps that don’t use AI — posting to Slack, creating the Jira ticket, sending the email — run free and unlimited. PagerDuty is one of 200+ tools Carly connects to; see the PagerDuty integration page and the full integrations list.
Side by side
| Claude + PagerDuty MCP | Carly | |
|---|---|---|
| Read on-call schedules & incident timelines | Yes | Yes |
| Create / update an incident | Yes (in chat) | Yes |
| Reacts the moment an incident fires | No | Yes, on triggers |
| Fans out to Slack + Jira + email automatically | No | Yes |
| Schedules the postmortem | No | Yes |
| Runs at 3am with your laptop shut | No | Yes (cloud) |
| Setup | Custom connector + paid Claude plan | Plain-English interview |
| Pricing | Paid Claude plan | AI agents from $35/mo |
Frequently Asked Questions
Does Claude integrate with PagerDuty?
Yes. PagerDuty publishes an official MCP server — a PagerDuty-hosted endpoint at mcp.pagerduty.com/mcp plus an open-source self-hosted build. Add it as a custom connector (paid Claude plan required) and Claude can read on-call schedules, list incidents, pull timelines, and create or update incidents inside a chat.
Can Claude create a PagerDuty incident?
Yes — the server is read/write, so Claude can create and update incidents and work with escalation policies. But it only does so inside a conversation you start; it can’t notice an incident firing and respond on its own. For that you need a trigger-based agent like Carly.
Will Claude alert my team when an incident fires?
No. Claude connectors have no event triggers, so a page at 3am sits unnoticed until you open a chat and ask. To auto-post to Slack, open a Jira ticket, and email stakeholders the moment an incident triggers, use an agent that runs on events — Carly.
Is there a PagerDuty plugin for Claude Code?
Yes. PagerDuty shipped a Claude Code plugin in Anthropic’s marketplace that scores uncommitted code changes against historical incident data. It’s a pre-deploy safety check for developers, not an on-call responder that fans out when a live incident fires.
What does automated PagerDuty fan-out cost with Carly?
AI agents start at $35/month, and the non-AI steps in the workflow — posting to Slack, creating the Jira ticket, sending stakeholder email — run free and unlimited, so a full incident pipeline stays affordable even during a noisy week.
More: Claude connectors · Can Claude send emails · Claude vs Carly · Claude Cowork alternatives · Best AI workflow automation tools · Claude + Databricks · Claude + Dialpad · Claude + Missive
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
