Paste one metadata record. The Weaver reads DCAT, schema.org, CKAN or an ODCS data contract, writes it out in all four, and shows exactly what each catalogue is missing, including the 13 critical data asset properties. It runs in your browser. Nothing is uploaded.
Don't paste personal data. Contact points should name a team or role, not a person.
| Property | Status | Why |
|---|
Does the DCAT output meet each profile? Must properties are required; should properties are recommended. These are simplified shapes that catch the common gaps: check against the official SHACL shapes (from SEMIC for the European profiles) before you publish. The UK profile uses the critical data asset properties from the GOV.UK metadata guidance, including its allowed values for access rights and classification. The UK GEMINI check lists what a geospatial record still needs (for example lineage, recorded as dct:provenance): for full GEMINI XML use your GIS catalogue.
Edit any field and the outputs update. Lists take commas. Distributions take one per line: title | url | format. Each field is mapped through the Metadata Rosetta.
| Field | Value | Rosetta concept | CDA property |
|---|
Before you publish a new version of an ODCS data contract, check it, and compare it with the last version. The check lists changes that would break the people who use your data, and the version number to use.
The Weaver is part of the open toolkit (MIT licence). Use it in your own catalogue, pipeline or AI agent, with no dependency on this site.
python -m dqkit weave record.json --report python -m dqkit weave record.json --to dcat python -m dqkit contract-check new.yaml --against old.yaml
from dqkit import weaver, contracts w = weaver.weave(record) w["outputs"]["dcat"], w["cda"]["coverage"] contracts.diff_odcs(old, new)
The toolkit's MCP server offers weave_metadata and check_contract, read-only. See the tutorial.
Fail a pull request when a contract change breaks consumers without a major version bump: contract-check exits with 1.
A crosswalk tells you that dct:title and name mean the same thing. Teams still have to retype records for each catalogue, and gaps only show up after publishing. A weaver does the translation, keeps one source of truth, and shows gaps before anything goes live.