Incidents: One Failure, One Thing to Work On
Platform alerts now become incidents: one record per failure with a timeline, acknowledgement, assignment and live updates, visible to the tenant who owns the cluster and reachable through the dashboard, GraphQL and MCP.