Profile Asset File

Reads an uploaded file and reports what is in it, without ingesting it.

Nothing is created: no ingestion task, no staged locations, no stored file. Use it to answer "what does this file actually contain, and can you place it" before committing to a run.

It returns no verdict. A row is described, never rejected, so there is no compliance judgement anywhere in the response.

Query parameters (all optional):

  • normalize_input โ€” run AI schema detection while loading. Defaults to true; without it the profile cannot report which columns carry what. Detection only runs for CSV/XLS/XLSX, so a GeoJSON or shapefile leaves field_coverage as undetermined regardless.
  • location_type โ€” facility | plot | shed. A hint, honoured when given, never required: every row is otherwise classified on its own evidence. See Location Types.
  • country, commodity_type โ€” context. Echoed under detected, and used to prefill eudr_payload only when they resolve.
  • limit โ€” max features returned (default 100, minimum 1). The aggregates under properties are always complete regardless of this.
  • include_eudr_profile โ€” adds the EUDR lens: the 4 ha band, the point_large_area kind, and eudr_payload. Off by default, so the core read never asserts the band.
  • area_band_ha โ€” override the band. Applied only under the lens.
  • include_null_geometry, cleanup_geometries โ€” govern geometry cleanup.

The response is a FeatureCollection with a non-standard top-level properties block carrying the file-level aggregates. That block is the point of the endpoint; features is per-row detail, best first, capped by limit.