What’s included
Each run writes one file per dataset, and each dataset has its own folder in the bucket:
Every file includes
exported_at, the UTC timestamp of that snapshot.
Supported formats: Parquet (default) and CSV.
Additional datasets are available on request.

