SigSentrySigSentry

Severity thresholds

How channels filter by severity, with tenant defaults, per-channel overrides, and personal preferences

Each notification channel has a severity threshold. The channel fires only for analyses whose severity is at or above that threshold. This is the primary lever for routing the right level of noise to the right place.

How "at or above" works

Analysis severity is ordered: critical > high > medium > low > info. The channel threshold has four valid values — critical, high, medium, low — and selects every analysis at or above that level.

A channel with threshold medium receives notifications for analyses whose severity is medium, high, or critical — but not low or info.

Channel thresholdReceives
criticalCritical only
highHigh and critical
mediumMedium, high, critical
lowLow, medium, high, critical (everything except info)

Routing examples

A common setup uses three Slack channels with different thresholds:

ChannelThresholdReceives
#oncall-criticalcriticalPages only
#engineering-incidentshighHigh-severity incidents
#engineering-noisylowMost analyses, including triages

Tenant default threshold

Under Settings → Notifications, set the default severity threshold for the whole organization. This is used:

  • As the default for newly created channels (you can override per channel)
  • For any channel that explicitly opts in to the tenant default

Most users set the tenant default to high and then override on specific noisy channels.

Per-user preferences

Each user has personal preferences under Settings → Notifications:

PreferenceDefaultWhat it does
Notify me on my analysesOnWhen you trigger an analysis (dashboard, API key with your identity), you receive a notification regardless of channel thresholds
Notify me only via emailOffSuppress chat notifications for analyses you triggered, only email

These are personal — they don't affect what other team members see in shared channels.

How thresholds combine in practice

When an analysis completes, every channel whose threshold is at or below its severity receives the notification. A high analysis fires channels with thresholds low, medium, or high — but not those with critical. Inactive channels are skipped.

If multiple channels qualify, all receive the notification — the #oncall-critical Slack might fire alongside an email to the on-call rotation.

Watchdog and severity

Watchdog rules generate analyses when their conditions trigger. The watchdog rule itself doesn't have a severity threshold — that's applied downstream by your channels. A watchdog rule that runs an analysis on every pattern match can produce many medium or low events; channels with threshold high will only see the spikes.

See Watchdog for details on rule configuration.

If you're not sure where to start:

ChannelThreshold
Tenant defaulthigh
Critical-only Slack channelcritical
Email to on-call rotahigh
Optional verbose Slack channellow

Iterate from there based on which alerts your team finds useful versus noisy.