Datasets with receipts
The Data Operations division of Blue Hen RE, as a product. Every dataset here carries its provenance (source, extractor, chunking strategy, and an OKF card), because training data you can't audit is training data you can't trust.
Catalog unreachable. Counts render when the production API responds (
GET /v1/catalog/stats).The pipeline
source → fetch (SSRF-guarded) → structure → chunk → OKF dataset card → catalog. Unchanged content is skipped by hash; every collection is logged; consented contributions enter the same pipeline with a provenance receipt.