5 Ingest Discovered Locations

Step 5 of 5 โ€” stage the discovered locations.

One call ingests the candidates of the discovery task from Steps 1-3. The candidates stay on the server: nothing is assembled in the client, and no file is uploaded. Because the coordinates were produced by discovery, they are staged exactly as found โ€” no geocoder runs over them again.

This replaces the older pattern of packing the candidates into a CSV and posting it to POST /ingestions/assets. That round-trip destroyed everything the CSV had no column for, and re-processed points the server had just produced.

What to send

location_type is the only required field. Omitting candidate_ids stages every candidate the default gate admits โ€” the usual choice. To stage a subset, send the id values from Step 3:

{
  "location_type": "facility",
  "candidate_ids": ["6c1eโ€ฆ", "8f2aโ€ฆ"]
}

Narrow the gate with filters when you want fewer candidates, e.g. {"filters": {"country": ["MY"], "min_confidence": "high"}}.

What comes back

A normal asset_task. Poll it at GET /ingestions/{asset_task_id} and wait for overall_status: staged โ€” an upload is finished at staged, not completed.

meta explains the selection: selected is how many were staged, filtered lists candidates the gate excluded (with reasons), and filters_applied shows the filters used including defaults.

If you name a candidate that the gate excludes, the call answers 422 with the reason rather than silently skipping it โ€” and creates no task.

Continuing the journey

{{asset_task_id}} and {{eudr_task_id}} are captured from the response, so the Facility / Supplier Ingestion and EUDR workflows carry on from here unchanged.

Required scope: ingestions:write