Skip to main content
The Alerts feed surfaces signals from your agents, scores, and adapter syncs. Each alert links to the move that caused it. Acknowledge to clear from the feed; act on the move that matters.

​
Alerts

The Alerts feed surfaces signals generated by your agents and scoring engine. Alerts notify you when something meaningful moves: a dimension drops below a threshold, an adapter sync detects a regression, or an agent surfaces a pattern worth your attention. Hit Alerts in the main nav to see all signals for your account.

​
Alert types

Score regression

A dimension or composite dropped meaningfully since the last re-score. Carries before/after values and the affected dimension.

Score improvement

A dimension lifted, typically after a team move or re-score. Receipt that something worked.

Threshold breach

A metric crossed a configured threshold (delivery velocity dropped below your target, for example). Generated by agent triggers set to event mode.

Adapter signal

An adapter sync detected an anomaly: unusual PR patterns, sprint velocity drop, customer signal spike, mapped to a specific dimension.

Agent finding

An agent run surfaced a finding that meets your notification criteria. Links to the full agent artifact.

Lifecycle transition

Your lifecycle placement moved based on recent self-assessment progress.

​
Reading an alert

Each alert carries:
FieldRead
TypeCategory of the signal (regression, improvement, threshold, others)
ProductWhich product the alert is about
DimensionThe specific dimension affected (if applicable)
DeltaThe move in value that triggered the alert
SourceWhether the alert came from a score, adapter sync, or agent run
TimeWhen the alert was generated

​
Acknowledging alerts

Alerts stay in your feed until acknowledged. Hit Acknowledge on any alert to mark it seen and move it to history. Acknowledging doesn’t resolve the underlying move; it just clears the active feed.
Use acknowledgment as a lightweight triage step. Acknowledge noise, act on signal.

​
Filtering

Use the filter bar to show:
  • Unacknowledged only for alerts waiting on you
  • By product to focus on a specific product
  • By type to see only regressions, improvements, or agent findings

​
Alert sources

SourceWhat creates it
Scoring engineEvery re-score compares against the previous run and generates delta alerts
Adapter syncDaily sync picks up anomalies in operational signal (GitHub, Linear, others)
Agent triggersEvent-based agent triggers fire alerts when conditions are met
Lifecycle assessmentStage transitions surface a notification

​
Alert retention

Alerts are retained for 90 days, then archived. Acknowledged alerts move to history immediately and stay searchable for the full retention window.

​
API access

Retrieve and acknowledge anomalies programmatically: 
# List unacknowledged anomalies (paginated)
GET /api/anomalies?severity=high&limit=25

# Acknowledge an anomaly
POST /api/anomalies/acknowledge
{ "anomalyId": "anm_..." }

# Convert an anomaly to a Linear issue
POST /api/anomalies/{id}/convert-to-linear
{ "teamId": "optional-linear-team-id" }

# Adjust detection sensitivity for an event type
POST /api/anomalies/sensitivity
{ "eventType": "pr_cycle_time", "threshold": 3.0 }
See List anomalies, Acknowledge anomaly, Convert to Linear, and Set sensitivity override. Legacy alert listing is available at GET /api/intelligence/alerts for backwards compatibility.

Agent Studio

Configure agents and triggers to generate targeted alerts.

Integrations

Connect tools to enrich adapter-sourced alerts.