Count runs per product
Total and failed runs for several products in one read, over the same window and traffic filter as GET /v1/runs/stats. A product counts its agent capabilities’ runs by agentId and its other capabilities’ runs by flowId, each family read the way /v1/runs/stats reads it, and the two summed. Name at most 100 products; one the caller does not own is left out, as is one that does not exist. An API key needs PRODUCTS:READ, plus AGENTS:READ or FLOWS:READ for each family the named products run.
Authentication
Query parameters
live keeps runs real users and integrations started; test keeps runs started from the dashboard (Test, step tests, Ask); eval keeps eval-batch runs; sim keeps simulation turns; all, or omitted, keeps every run. A simulation turn is sim even when a schedule started it, a schedule run is live, and an eval batch run is eval whoever started it.
Window start (ISO 8601). Defaults to 24 hours ago.
Window end (ISO 8601). Defaults to now.
Comma-separated product ids, at most 100. A product the caller does not own is left out of the answer.
Response
The started_at window the query resolved to, after defaults.