Skip to main content
Alert rules let you define conditions that trigger notifications when your workflows cross a threshold. Use them to stay informed about failure spikes, callback issues, or unexpected credit consumption.
Alert rules watch rates across a workflow and notify you by email or in-app. To act on a single failed run instead, for example by posting it to an external system, give the workflow an error handler.

​
Creating a rule

  1. Go to Settings → Alert rules
  2. Click New Alert Rule
  3. Configure the rule (see below)
  4. Click Create Alert Rule

​
Rule configuration

​
Available metrics

Rate metrics are calculated over the window you choose in Measured over and expressed as a value between 0 and 1. A threshold of 0.15 means the alert fires when the rate exceeds 15%.
Credit Balance is an organization-wide metric. Pair it with the Less Than operator to get notified when your remaining credits drop below a threshold so you can top up before runs start failing.

​
Notification channels

  • In-App: alerts appear in real time via the notification area in the app
  • Email: alerts are sent to the specified email recipients
Select both to have a breach reach the app and the mailbox at once. Each channel delivers independently, so a mail delivery problem never suppresses the in-app notification.

​
Workflow scope

By default, a rule evaluates metrics across all workflows. To monitor a specific workflow, select it in the Scope field when creating or editing the rule. The Credit Balance metric is always organization-wide and cannot be scoped to a workflow.

​
Managing rules

From Settings → Alert rules, you can:
  • Edit: update any rule configuration
  • Enable/disable: toggle a rule without deleting it
  • Delete: permanently remove a rule

​
How a rule is evaluated

Every enabled rule is checked once a minute. A rule does not notify you the instant it crosses its threshold, because a metric that sits near its threshold would otherwise alternate between triggered and resolved indefinitely. Instead it moves through four states: Only two moves notify you: Alert Triggered when a rule goes from Pending to Firing, and Alert Resolved when it goes from Recovering back to OK. A brief spike that clears before the pending period elapses never reaches you at all, and a brief relapse during a recovery does not produce a second Alert Triggered for the same incident. The Breaching since line on an alert email dates from when the condition actually started, not from when the pending period expired. If you have set a repeat interval, reminders are sent only while a rule is Firing.
An empty window counts as being within the threshold, as does a rate window holding fewer than five readings, which is too little evidence to trust a percentage. Both feed the recovery timer, which is what lets a firing rule resolve on its own when a workflow stops running. A rule whose metric genuinely cannot be measured holds its current state instead.

​
Permissions

Managing alert rules requires a role with Core.AlertRule permissions: