What changed

Adds docs/external-rdf-generators.md — a reference documenting the exact RDF/XML profile parseRDF() (src/lib/rdf/parser.ts) accepts — plus a row in the README documentation table.

It covers, for authors of external tools that want to emit Playground-compatible ontologies programmatically:

• how entity ids are derived from class URIs (uncapitalized local name)
• property attachment via exact rdfs:domain URI match, ont:propertyType values, and the XSD range fallback mapping
• relationship endpoints (ont:fromEntityId/ont:toEntityId vs domain/range), valid cardinalities, and relationship attributes via ont:relationshipAttributeOf
• display annotations (ont:icon, ont:color, ont:unit, ont:enumValues, ont:isIdentifier)
• constructs the parser deliberately ignores (e.g. rdfs:subClassOf), with the recommended alternative
• a minimal complete example and verification tips (round-trip via the serializer, headless parseRDF() in CI)

Why

The existing Authoring Guide targets humans writing ontologies by hand. While integrating two external Python tools (a schema.org schema generator and a knowledge-object framework) with the Playground, I had to reverse-engineer these rules from parser.ts. Documenting the import profile makes the Playground usable as a visualization front-end for external pipelines — generators can emit .rdf artifacts that import cleanly via Import/Export, the catalogue, and the embed widget, without reading the parser source.

Notes

• Docs-only change; no code touched.
• The behaviors documented were verified against a live dev instance by importing programmatically generated files (entity ids, property types, cardinalities, stub handling all render as described).

🤖 Generated with Claude Code