Contribute

Contributions must be factual and sourced. The repository contains the data, the schemas and the methodology; the website is generated from the same source. Every accepted change ships in the next dataset release and is recorded in the changelog.

Three issue templates are available: correction, dataset request and schema proposal. Pick the closest one — maintainers will reroute if needed.

Pull requests must pass schema validation. CI runs the validator on every changed record. PR descriptions should cite at least one source URL.

Found an incorrect field? A correction needs the record, the field, the proposed value and a source. The fastest path is the correction form, which opens a pre-filled issue.

Registrars may submit a capability matrix as a pull request. Submissions are merged with the verification statusregistrar_submitted. Once an independent check confirms the record, the status updates to registrar_verified.

Schema changes are proposed as pull requests against the schema file and discussed in an issue first. New fields must be optional until backfilled, and every enum value needs a definition. Accepted changes are recorded in the changelog with the release they ship in.

A field becomes independently_tested only when an automated probe confirms the documented behaviour against the live service. Contributors do not need to run these probes themselves — maintainers do so on a daily cadence.

The project follows the Contributor Covenant. Read the full text in CODE_OF_CONDUCT.md.