Skip to main content
Alerts evaluate metric conditions when a run completes and notify a channel (Slack, email, webhook, or a Human Review project) when they trigger. The CLI command is coval monitors; the product and the API call these alerts, and the command name is kept for compatibility. create and update also accept --input-json (inline JSON, @file, or - for stdin); explicit flags override fields from the JSON. See Input JSON.

List Alerts

Output columns: ID, Name, Status, Scope, Evaluation, Triggers

Get Alert

Create Alert

Conditions

Each condition is an object with a metric_id, an aggregation (SINGLE, RUN_AVERAGE, or RUN_FRACTION), an operator (GT, GTE, LT, LTE, EQ, NEQ), and a threshold (threshold_float or threshold_string). Fraction conditions can add match_value and match_mode (exact, contains, contains_case_insensitive); rolling windows use window_size_days or window_size_runs.

Channels

Each channel is an object with a channel_type (SLACK, EMAIL, WEBHOOK, or HUMAN_REVIEW) and a config object, for example {"recipients": ["team@company.com"]} for EMAIL or {"channel_id": "C0123ABC", "channel_name": "#alerts"} for SLACK.

Update Alert

Delete Alert

Test an Alert Against a Run

Evaluate an alert’s conditions against a completed run without sending notifications.

Alert Events

List the evaluations recorded for an alert.
Output columns: Event ID, Monitor ID, Run ID, Outcome