Project ID of the project you're trying to access. To find the ID of the project, make a call to /api/projects/.

UUID of the evaluation this report config belongs to.

How report generation is triggered. 'every_n' fires once N new evaluation results have accumulated (subject to cooldown_minutes and daily_run_cap). 'scheduled' fires on the cadence defined by rrule. * `scheduled` - Scheduled * `every_n` - Every N

RFC 5545 recurrence rule string for scheduled reports. Only daily and weekly cadences are supported: use 'FREQ=DAILY' or 'FREQ=WEEKLY;BYDAY=MO,FR'. Required when frequency is 'scheduled'; ignored otherwise.

Read-only anchor datetime used to expand scheduled reports. The server sets this automatically when a report is switched to scheduled mode.

Read-only timezone used for scheduled reports. Evaluation reports use UTC.

List of delivery targets. Each entry is either {type: 'email', value: 'user@example.com'} or {type: 'slack', integration_id: <int>, channel: '<channel>'}. Slack integration_id must belong to this team.

Maximum number of evaluation runs included in each report. Defaults to 200.

Whether report delivery is active. Disabled configs do not fire.

Read-only. Report configs are soft-deleted only when their evaluation is deleted. Use enabled=false to stop deliveries.

Optional custom instructions appended to the AI report prompt to steer focus, scope, or section choices without modifying the base prompt.

Number of new evaluation results that triggers a report (every_n mode only). Min 100, max 10000. Defaults to 100. Required when frequency is 'every_n'.

Minimum minutes between count-triggered reports to prevent spam (every_n mode only). Min 60, max 1440 (24 hours). Defaults to 60.

Maximum count-triggered report runs per calendar day (UTC). Min 1, max 24 (one per cooldown window). Defaults to 10.