Features

Alerts

Get notified when detections fire, metrics cross thresholds, or ingestion stops.

Detections annotate traces after the fact. Alerts push the important ones to you in real time. Configure alert rules under Settings → Alerts to be notified when something you care about happens — a detection spikes, a cost or latency metric crosses a threshold, or trace ingestion goes quiet.


Rule types

TypeFires when
ThresholdA single detection occurs more than N times within a time window
Per-eventA detection fires at all (every matching event)
CompoundMultiple detection conditions combine with AND / OR logic
Metric thresholdAn aggregated metric (e.g. cost, latency, token count) crosses a value
Ingestion healthTrace ingestion stops or drops unexpectedly

Each rule has a severity (critical, warning, or info), a window over which it evaluates, and a cooldown so you aren't paged repeatedly for the same ongoing condition.


Creating an alert from a detection

The quickest path is from a Detection: open it and create an alert rule directly, pre-scoped to that detection. From there, set the threshold count, window, severity, and where the alert should go.


Destinations

Route alerts to Slack (connect it under Integrations), email, or surface them as in-app notifications. Webhooks let you deliver alerts to any custom endpoint.

Alert rules are evaluated server-side against incoming spans — they don't affect how your application runs, and adding or removing a rule never touches your code.

On this page