Library Records

Query library records by dataset, key prefix, and date in force, and upsert up to 2000 records with their source URL and quote through /v1/data-library/records.

GET /v1/data-library/records returns the records of the shared library and of the workspace's own datasets, ordered by dataset key, record key and start date. Three filters narrow the read. dataset matches one dataset key exactly. prefix matches the start of the dataset key, which reads a whole family such as every dataset under one prefix, and record matches the start of the record key. %, _ and \ in prefix and record match literally, so a key containing an underscore is found by that underscore and not by any character.

on keeps the records in force on a date: a record is returned when its valid_from is empty or not after that date, and its valid_to is empty or not before it. An open end therefore always matches. A value that is not YYYY-MM-DD is ignored, and the read behaves as if on were absent. limit defaults to 1000 and is clamped to 1 through 5000. Each record carries shared, its dataset and record key, the free-form attributes, its validity period, source_url, quote and retrieved_at.

curl "https://api.talonic.com/v1/data-library/records?prefix=acme.&on=2026-03-15&limit=200" \
  -H "Authorization: Bearer $TALONIC_API_KEY"

Add or Replace Records

POST /v1/data-library/records stores up to 2000 records under one dataset for the workspace of the key and answers 201 with the number stored. A record with the same record key and valid_from as a stored record of the workspace replaces it, including its attributes, validity end, source and quote. A record whose valid_from differs is stored next to it as another period. Entries with an empty record_key are skipped and not counted in stored. valid_from and valid_to must be YYYY-MM-DD or null; any other format is refused.

Keep the source on every record. source_url and quote are what let a reader of the library, or an app that prices against it, verify a value against the publisher. The quote is the publisher's own wording, up to 4000 characters, copied from the document. A body with more than 2000 records, a record without record_key or attributes, or a malformed field is refused with 400 before anything is stored.

curl -X POST "https://api.talonic.com/v1/data-library/records" \
  -H "Authorization: Bearer $TALONIC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "dataset": "acme.heat-tariff",
    "records": [
      { "record_key": "base-price", "attributes": { "eur_per_mwh": 98.4 }, "valid_from": "2026-01-01", "valid_to": "2026-12-31", "source_url": "https://example.com/tariffs/heat-2026.pdf", "quote": "Arbeitspreis 98,40 EUR/MWh" },
      { "record_key": "base-price", "attributes": { "eur_per_mwh": 101.2 }, "valid_from": "2027-01-01", "source_url": "https://example.com/tariffs/heat-2027.pdf", "quote": "Arbeitspreis 101,20 EUR/MWh" }
    ]
  }'

{ "data": { "stored": 2 } }