usage

View Scenario API usage statistics for a date range. Default response is summarized (~1–3 KB regardless of activity volume): headline CU totals, per-user consumption, aggregated per-model CU/jobs (sorted desc, joined with model names), and per-asset-kind counts. Heavy sections are stubbed with counts and opt-in hints — pass them in `include` to expand. Args: - start_date, end_date: optional ISO dates to bound the query - team_id, project_id: optional context filters - include: optional array to expand specific sections. Accepted values: • 'usages' — per-usage-type aggregated totals (images-generation, training, custom, ...) • 'usages.daily' — full per-day timeseries per usage type • 'modelUsages.daily' — full per-day per-model timeseries • 'nsfwUsages' — content-moderation totals (non-zero only) • 'activity' — raw per-event activity log • 'entities' — denormalized lookups for ids referenced above (avatar URLs and full job inputs are stripped) - response_format: 'json' (default) or 'markdown' Returns: { totals, consumption, modelUsages, assetUsages, sections, _hint } by default. Each value in `include` adds the corresponding section to the response. Examples: - "Show usage for this month" -> start_date="2026-04-01", end_date="2026-04-30" - "What models cost me the most?" -> default response (modelUsages is already aggregated and sorted) - "Break down model CU by day" -> include=['modelUsages.daily'] - "Give me the raw event log" -> include=['activity'] Don't use when: You need per-job execution state or cancellation. Prefer manage_jobs instead.

{
  "team_id": "team_abc123",
  "project_id": "proj_xyz789",
  "start_date": "2026-03-01",
  "end_date": "2026-03-31"
}

{
  "usages": [
    {
      "date": "2026-03-20",
      "credits": 120
    }
  ]
}