The DataCite Metadata Schema 4.7 presents metadata not as a secondary administrative layer, but as the epistemic infrastructure through which research outputs become citable, discoverable and reusable across disciplinary boundaries. Its central ambition is pragmatic universality: the schema is deliberately discipline-agnostic, suitable for datasets, software, images, samples, instruments, preprints and other research resources, while remaining focused on accurate identification rather than replacing richer community-specific description. At its core are six mandatory properties—Identifier, Creator, Title, Publisher, PublicationYear and ResourceType—which together generate the minimal architecture of citation. Yet the schema’s intellectual value lies in its recommended extensions: Subject, Contributor, Date, RelatedIdentifier, Description and GeoLocation enlarge the resource from an isolated object into a relational scholarly entity. Version 4.7 strengthens this relational logic by adding Poster and Presentation to resourceTypeGeneral, RAiD and SWHID to relatedIdentifierType, and “Other” as a relationType, while introducing relationTypeInformation for more precise contextualisation. As a case study, a dataset can be rendered not merely as a file with a DOI, but as a situated research object: authored through ORCID-linked creators, published by a ROR-identified repository, licensed through standardised rights metadata, connected to cited papers, methods, funding and spatial coverage. The conclusion is clear: DataCite’s schema is a grammar of scholarly persistence, converting research outputs into durable nodes within an interoperable knowledge commons.