The Alerts API connects external monitoring systems to Prometheux. An alert carries project_id and concept labels; when it fires, the platform stores it and can automatically run the targeted concept. You can then browse the alert history and reprocess any alert to re-run its concept on demand. All paths are relative to the base URL and require authentication. All responses use the standard envelope.

Webhooks

Point your monitoring system’s webhook receiver at these endpoints. Each alert rule should set two labels — project_id (the Prometheux project) and concept (the concept to run when the alert fires).
Use POST /alerts/webhook/test to verify connectivity from your monitoring system. It echoes back the payload and metadata without storing an alert or running a concept.

Alert history

Return stored alerts, most recent first, with offset/limit pagination.
The alert history table is created lazily the first time an alert is stored. If you call /alerts/history on a workspace that has never received an alert, you may see an error until the first webhook arrives.

Reprocess an alert

Re-run the concept associated with a stored alert — useful for backfilling after fixing a concept or compute issue.
Returns 404 if no alert with that ID exists in the given scope.