GBIF Dataset Endpoint & License Monitor

Watch one registered GBIF dataset for exact endpoint URL/type changes, additions/removals, license changes, publisher changes, DOI changes, title/type changes and a logical deletion timestamp. Built for biodiversity data pipelines and research data inventories that need a durable before/after record rather than another occurrence search.

Input

{"datasetKey":"50c9509d-22c7-4a22-a47d-8c48425ef4a7","monitorName":"default","includeUnchanged":true}

Use the UUID from a GBIF dataset page. Keep monitorName stable; a different name starts a separate baseline. Save as an Apify Task and schedule it to monitor repeatedly.

Output

Every successful check writes OUTPUT and by default one dataset row: BASELINE, CHANGED or UNCHANGED, with checkedAt, previousCheckedAt, changes, current and previous snapshots. Endpoint identity uses GBIF endpoint keys; ordering does not create changes. Warnings flag DATASET_MARKED_DELETED, LICENSE_UNSPECIFIED and NO_REGISTERED_ENDPOINTS. No contacts, emails, occurrence locations or full dataset files are collected.

GBIF logical deletion is a metadata timestamp; a 404, incomplete response or malformed endpoint fails the check and preserves prior history. A registered endpoint is not proof its download works: endpoint URLs are recorded but never fetched. Dataset license metadata is not a legal assessment of individual occurrence/media rights. Review source terms and citation requirements.

Pricing

$0.01 per successful dataset check, including baseline and unchanged checks. Platform usage included. Failed checks are not charged. includeUnchanged=false suppresses unchanged dataset rows but OUTPUT is still written and the check is still charged. Set maximum charge to at least $0.01.

Reliability and limits

Fixed official api.gbif.org source only, one UUID, 2MB response cap, 20-second deadline, no redirects or retries, at most1000 endpoints. Defaults256MB/60seconds; restart and standby off. No API key, paid proxy or external account required. History is isolated by user, Actor, dataset and monitor. A best-effort five-minute lease rejects overlap; it is not an atomic storage lock, so schedule each monitor sequentially. Output is delivered before charging; history commits only after a successful charge. Interrupted output/charge/commit can cause repeated output or charges on retry. Storage retention or deleting history starts a new baseline.

Sources and support

Official API: https://techdocs.gbif.org/en/openapi/v1/registry-principal-methods Citation: https://www.gbif.org/citation-guidelines QuietDataTools is not affiliated with GBIF. Use the official free API directly if you do not need stateful monitoring. No revenue or demand guarantee. Report the dataset UUID and run ID in Issues without credentials.