Alerts and notifications
Logging an event and alerting about it are deliberately separated. The API stores the event and returns instantly; a background job sends notifications within a few minutes. A slow email provider can never slow down, or break, the app doing the reporting.
Who gets told, and why
When an error arrives, the engine checks two things:
| Alert rules | Set by admins for the organisation: "app X (or all
apps), integration Z, class Y, severity S or worse goes to this address on
this channel". This is how IT hears about technical faults regardless of
who is subscribed personally. The integration filter is what lets one app's
flows go to different people: name a flow exactly, or use * as a
wildcard, e.g. *salesforce* for every Salesforce flow. |
|---|---|
| Personal subscriptions | Chosen by each user in My alerts: the apps, integration, error class and minimum severity they care about, delivered instantly or in the daily digest. See Using the admin area. |
Successes never trigger alerts; they feed the dashboard and reports.
Flood safety
A flapping integration cannot bury people in email:
- Critical errors always send instantly to instant subscribers.
- Lower severities are throttled: after one alert, the same rule or subscription goes quiet for a window (15 minutes by default) and further matching events are recorded as throttled rather than sent.
- The daily digest (8am) rolls up everything a digest subscriber's choices matched since yesterday, one email, deduplicated, capped at 30 rows.
- Every alert attempt is kept in an audit trail on the event's detail page, including failures and throttles, so "did anyone get told?" always has an answer.
Channels
| Live. Every notification says why you received it and links to the event. One email per person per event, however many subscriptions match. | |
| Teams / Slack | Live. Point an alert rule at a channel's incoming webhook URL and matching errors post straight into the channel. |
| SMS / WhatsApp | Built into the schema and rules UI but awaiting a provider decision; rules on these channels are flagged as stubs and queue rather than send. |
| Salesforce push | Future-proofed: business errors will be pushable
into Salesforce as well as pollable via GET /events today. Design
pending. |
Next: Using the admin area