The rijksmuseum-mcp+ dataset is an enriched copy of the Rijksmuseum's CC0 Linked Open Data, harvested in May 2026 from the museum's Linked Art resolver, OAI-PMH feed and Search API. It covers about 834,000 artworks and 418,000 vocabulary terms (people, places, materials, techniques, Iconclass subjects), connected by nearly 15 million links between artworks and terms, and its entities carry identifiers that link out to Wikidata, VIAF, ULAN and RKD. It was enriched by parsing the free-text provenance records into structured ownership events (owners, transfer types, dates, prices) aligned with the CMOA/PLOD provenance models, with a small part of that work done by LLMs. Also parsed are the inscriptions and collectors' marks, normalised dates and dimensions so they can be queried as ranges, and place coordinates where they come from Rijksmuseum authority records.

N.B. This is an unofficial dataset that the museum has neither reviewed nor endorsed. It is available as a single SQLite file (527 MB compressed): https://github.com/kintopp/rijksmuseum-mcp-plus/releases/download/v0.90/vocabulary.db.gz You can also explore and query the dataset via a demo MCP server: https://github.com/kintopp/rijksmuseum-mcp-plus/