Pagerduty

Dedalus MCP server for PagerDuty incident response. It exposes incident reads, incident acknowledgement and resolution, on-call lookup, and service listing.

Tools

• list_incidents - List incidents with status, service, and time filters.
• get_incident - Fetch one incident by ID.
• acknowledge_incident - Acknowledge an incident.
• resolve_incident - Resolve an incident.
• list_oncalls - List on-call entries.
• list_services - List PagerDuty services.

Safety Notes

acknowledge_incident and resolve_incident are high-impact incident-response writes. Acknowledging may change escalation behavior. Resolving may stop paging and incident follow-up. Require explicit user intent and a specific incident ID before calling either tool.

PagerDuty writes require a From email for auditability. Pass from_email or set PAGERDUTY_FROM_EMAIL.

Never commit API tokens, incident exports, escalation policy data, or user schedules.

Authentication

This implementation targets Dedalus Type 3 DAuth API-token auth. It declares a connection named pagerduty-mcp with secret key PAGERDUTY_API_TOKEN and sends Authorization: Token token=<token> through the DAuth dispatch layer.

Environment variables:

PAGERDUTY_API_TOKEN=...
[email protected]
DEDALUS_AS_URL=https://as.dedaluslabs.ai

If your PagerDuty OAuth app is ready, configure OAuth variables in Dedalus and keep the same bearer-compatible connection model only after verifying the token header expected by your app.

Run Locally

uv run python src/main.py

In another terminal:

MCP_SERVER_URL=http://127.0.0.1:8080/mcp uv run python src/_client.py

Set PAGERDUTY_TEST_INCIDENT_ID to a safe test incident. The client intentionally calls the write tools, so do not point it at a production incident unless that is the intended test.

Deploy

Publish as dedalus-labs/pagerduty-mcp, configure PAGERDUTY_API_TOKEN as a required DAuth credential, deploy, then run:

DEDALUS_MCP_SERVER_ONLY=1 DEDALUS_MCP_SERVER=dedalus-labs/pagerduty-mcp uv run python src/_client.py