The moment you know what a runbook should say is halfway through the incident it would have prevented, and that is exactly when nobody has time to write one. Having create_runbook reachable from the same conversation where the alert is being triaged is the only way that habit survives. Be deliberate with manage_alert_definitions though — it deletes, and an agent removing an alert definition is a silent failure that only shows up the next time something breaks.
Doctor Droid's AIOps endpoint. Twelve tools that split into reading the incident picture, reading configuration, and changing both.
- search_alerts — across tools and channels, with severity, source and time breakdowns
- search_resources — dashboards, services and infrastructure across connected platforms
- get_service_catalog — services with their owners, dependencies and metadata
- list_investigations and list_runbooks — recent AI investigations, and the runbooks and wiki documents
- update_service, create_runbook, update_runbook — write the catalogue and the runbooks
- view_alert_definitions and view_auto_investigation_rules — the configuration, with sample alerts and DSL
- manage_alert_definitions and manage_auto_investigation_rules — create, update and delete them
A bearer API key; there is no OAuth. The Business plan offers self-hosting inside your own VPC with no outbound traffic to Doctor Droid.
Paste a URL with a key — add the endpoint to your client