Alert Inbox Triage #

The Alert Inbox (/v1/platform/alerts) aggregates deduplicated infrastructure and mail pipeline alerts without exposing tenant PII or message payloads.

Alert Lifecycle & Actions #

StateMeaningOperator Action
openActive unacknowledged conditionInvestigate root cause; acknowledge to claim ownership
acknowledgedOperator actively investigatingContinue remediation or snooze during scheduled work
snoozedNotification noise temporarily mutedAuto-reopens when snooze timer expires unless resolved
resolvedUnderlying fault verified clearedCloses incident with resolution note

Health Impact Invariant #

IMPORTANT

Snoozing an alert only silences notification noise; it NEVER marks a degraded service healthy. The Service Topology health indicator continues to report degraded or unhealthy until the underlying probe passes and the alert is resolved.

API Operations #

  • List Alerts: GET /v1/platform/alerts (supports filtering and pagination).
  • Inspect Alert: GET /v1/platform/alerts/:id.
  • Acknowledge / Snooze / Resolve:
  • POST /v1/platform/alerts/:id/acknowledge
  • POST /v1/platform/alerts/:id/snooze
  • POST /v1/platform/alerts/:id/resolve

All mutating alert actions require platform.alerts.mutate, a unique request key (platformAlertsRequestKeyPattern), and are recorded in the platform audit log.