Prometheus Integration
Ingest Alertmanager firing alerts for Quality and MTTR metrics.
Why connect Prometheus
Prometheus (via Alertmanager) is a Monitoring & Observability source. After you connect, the platform reads firing and resolved alerts on a schedule and uses them for Quality metrics such as incident volume and mean time to recovery (MTTR).
The integration is read-only. We never change alert rules, silences, or write metrics into your Prometheus stack.
This connects to your Alertmanager (and optionally Prometheus) endpoint, not the platform’s own telemetry.
What you need
- A publicly reachable Alertmanager HTTPS base URL (required for alert ingest). Example:
https://alertmanager.example.com. - Optional Prometheus URL if you also want a health check against Prometheus
/-/healthy. - Optional Bearer token or HTTP basic credentials if your endpoint requires auth.
Alertmanager URL is required. A Prometheus URL alone is not enough to connect or sync alerts.
Localhost and private IP addresses are not supported from production. Use a publicly reachable HTTPS URL (or an endpoint the platform can reach through your allowlist).
Select Prometheus in your tech stack under Monitoring & Observability if you use it today.
Setup
- Identify your Alertmanager base URL (Alertmanager v2 API at
/api/v2/alerts). - In the platform, go to Settings → Integrations and choose Connect on Prometheus.
- Enter Alertmanager URL (required). Optionally add Prometheus URL and auth.
- Click Test connection. Success confirms Alertmanager (and Prometheus health when a Prometheus URL is set).
- Click Connect (or Save) to store encrypted credentials.
- Optionally click Sync now for an immediate pull, or wait for the daily batch sync.
Data we read
| Source data | Platform use |
|---|---|
| Active Alertmanager alerts (firing / resolved in window) | incident.declared / incident.resolved → Quality, MTTR |
Alert labels (severity, alertname, service) |
Incident metadata and service attribution |
We fetch alerts overlapping roughly the last 30 days on each sync (scheduled daily or Sync now).
Data we do not read or change
| Not accessed | Notes |
|---|---|
| PromQL queries or metric time series | No TSDB export |
| Alert rule definitions | Alerts only, not rule files |
| Silences or inhibition config (writes) | Read-only API |
| Pushgateway or remote-write targets | Not used |
Grafana is a separate connector. Use it when Unified Alerting and deploy annotations live in Grafana instead of (or in addition to) Alertmanager.
Sync schedule
After connect, the platform syncs on a daily batch schedule. Use Sync now in Integrations → Prometheus for an immediate pull. You receive an in-app notification when a sync completes or fails.
SDLC Engineering Health (Insights)
Under Insights → Engineering Health, Prometheus/Alertmanager sync contributes to the same weekly dashboard buckets as other monitoring sources (for example Time to Restore Service from resolved alerts).
Disconnect
Remove Prometheus under Integrations → Disconnect. Polling stops; previously ingested events remain in historical metrics until retention applies.