Skip to main content
Causeloop keeps you informed through two complementary systems: a notifications feed that logs events relevant to you, and alert rules that proactively push signals to Slack, PagerDuty, or email when risk crosses a threshold you define.
This page covers the notifications feed and alert rule management. For notification delivery preferences (which channels you receive notifications on, and for which event types), see Settings → Notifications.

Notifications feed

Viewing your notifications

Click the bell icon in the top bar to open the notifications panel. The bell displays an unread badge count that updates in real time. The feed is available via GET /v1/me/notifications/feed. It supports: Each notification item contains:

Notification types

Notification rows are materialized from real product events, fanned out to one row per active workspace member — never a demo placeholder. Every notification you see maps to something that actually happened:

Marking notifications read

Mark a single notification read Click the notification row to open the linked entity. The notification is marked read automatically. You can also call POST /v1/me/notifications/feed/{id}/read directly. Mark all notifications read Click Mark all read in the notifications panel header, or call:
To clear only a specific category, pass a type prefix:
This returns { "marked": 8, "unread_count": 4 } — the count of items marked and the remaining unread count. Dismiss a notification Click the dismiss (×) button on a notification row, or call DELETE /v1/me/notifications/feed/{id}.

Unread count

The unread badge is driven by GET /v1/me/notifications/unread-count. The Dashboard bell icon uses the same endpoint to show the badge dot.

Alert rules

Alert rules let you define conditions on predictions or issues that trigger proactive notifications — before an incident occurs.

The Alert rules panel

Alert rules appear on the Predictions screen in a sidebar panel. Each rule shows:
  • Name — a human-readable label (e.g. “Critical pattern alert”)
  • Condition summary — threshold, window, and delivery channels (e.g. ≥85% / 3D · SLACK + PAGERDUTY)
  • Enabled toggle — click to enable or disable the rule without deleting it

Creating an alert rule

1

Open the Predictions screen

Navigate to Predictions in the left sidebar.
2

Click New alert rule

The rule creation dialog opens. (Alternatively, use the API directly — see below.)
3

Set the rule name

Enter a descriptive name, such as “Inventory N+1 watch”.
4

Define the scope

Choose which patterns or entity types this rule watches. Examples:
  • A specific pattern_id to watch one pattern
  • A workspace-wide rule that watches all predictions
5

Set the threshold

Enter the probability threshold (0.0–1.0) and the prediction window in days (≥ 1). The rule fires when any in-scope prediction exceeds this probability within the window.
6

Choose delivery channels

Select one or more channels: email, slack, pagerduty, or in_app. For Slack, configure the channel ID in your integration settings. For PagerDuty, configure the service key.
7

Save

The rule is active immediately. You can test it against live data using POST /v1/alert-rules/{id}/test.
The rule is created via POST /v1/alert-rules with a body such as:

Managing alert rules

Snoozing a rule

To silence an alert rule temporarily without deleting it:
The rule resumes firing automatically after snoozed_until. To un-snooze early, send snoozed_until: null.

Alert rule fields

API reference

Notifications Alert rules

Predictions

Alert rules fire on prediction thresholds — review the predictions screen.

Search & Activity

The activity feed shows all fired alerts as events.

Settings — Notifications

Configure which channels you receive notifications on and for which event types.