List Record Sets

List record sets in the Talonic value plane with cursor pagination. Filter by layer or kind to find the tables your structuring and resolution runs wrote.

A record set is a table-like collection of records at one layer of the Talonic value plane, the typed cell storage that holds every structured value the pipeline produces. Three layers form the refinement ladder: structured (field values as extraction wrote them), resolved (canonical values produced by resolution runs), and product (final assembled output rows). Within a set, each record is a row pointer that maps to a source document, and each cell holds one field value with its own status, confidence, and provenance pointers back to the cell it was derived from. This endpoint lists the record sets in your workspace with cursor-based pagination.

Filter by layer to find sets at a specific pipeline stage, or by kind to locate sets from a particular origin: structuring_run for extraction output, resolution_run for standalone resolution output, and assembly_product or data_product for assembled product tables. Each entry carries lifecycle metadata (name, layer, kind, status, record and field counts) plus links to the detail, fields, records, and export endpoints. Cell values themselves are read through the records endpoint, not this list. An unrecognized layer or kind value is not an error — it simply matches nothing and returns an empty page.

Record sets are created by the runs that write them, and creation is idempotent per source: the platform keeps one active record set per source run and layer, so re-running, retrying, or resuming the same job writes into the existing set instead of minting a duplicate. Historical duplicates from before this rule were kept but marked superseded, and they still appear in this list with that status. Treat the newest active set for a given source_id as the authoritative table.

Pagination is cursor-based — a keyset over (created_at, id) — which stays stable while new sets are being created, unlike page-number pagination: limit accepts 1-100 (default 20), order sorts by creation date (desc by default), and each page returns a next_cursor until has_more is false. pagination.total counts every set matching your filters, not just the current page, so you can size progress indicators up front.

The record_count and field_count on each entry are cached summary columns maintained by the value-plane writers, so listing stays fast even for large workspaces. They are advisory: while a pipeline is actively writing, a set can briefly report counts behind the cells already committed. For exact, visibility-filtered numbers, read GET /v1/record-sets/{id}/records — its pagination.total counts the rows your key can actually see.

Response

curl -s "https://api.talonic.com/v1/record-sets?layer=resolved&limit=10" \
  -H "Authorization: Bearer tlnc_your_api_key"

{
  "data": [
    {
      "id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
      "name": "Resolution Run 2024-10-15",
      "layer": "resolved",
      "kind": "resolution_run",
      "source_id": "b2c3d4e5-f6a7-8901-bcde-f23456789012",
      "status": "active",
      "record_count": 142,
      "field_count": 12,
      "created_at": "2024-10-15T11:30:00.000Z",
      "links": {
        "self": "/v1/record-sets/a1b2c3d4-e5f6-7890-abcd-ef1234567890",
        "fields": "/v1/record-sets/a1b2c3d4-e5f6-7890-abcd-ef1234567890/fields",
        "records": "/v1/record-sets/a1b2c3d4-e5f6-7890-abcd-ef1234567890/records",
        "export": "/v1/record-sets/a1b2c3d4-e5f6-7890-abcd-ef1234567890/export"
      }
    }
  ],
  "pagination": {
    "total": 8,
    "limit": 10,
    "has_more": false,
    "next_cursor": null
  }
}

curl -s "https://api.talonic.com/v1/record-sets?layer=structured&kind=structuring_run&limit=10&cursor=YTFiMmMzZDQtZTVmNi03ODkwLWFiY2QtZWYxMjM0NTY3ODkwfDIwMjQtMTAtMTVUMTE6MzA6MDAuMDAwWg" \
  -H "Authorization: Bearer tlnc_your_api_key"

Errors