AnomalyDetector

Source: src/AWS/AMP/AnomalyDetector.ts

A Random Cut Forest anomaly detector inside an Amazon Managed Service for Prometheus workspace — continuously evaluates a PromQL query and emits anomaly scores as new metrics in the same workspace.

Creating an Anomaly Detector

const workspace = yield* AMP.Workspace("Metrics", {});
const detector = yield* AMP.AnomalyDetector("RequestSpikes", {
  workspaceId: workspace.workspaceId,
  alias: "request-spikes",
  query: 'rate(http_requests_total{job="api"}[5m])',
  evaluationInterval: "1 minute",
  missingDataAction: { skip: true },
});