AI skill: serialize the current UNS to an RDF/OWL graph file via the Export Tags format selector or the export_graph_model MCP tool.

AI Integration → Platform Skills Library → Export UNS as RDF

When to invoke

• “Export my UNS as RDF/JSON so I send it to the ontology group”
• “Save the solution as JSON-LD for our PIM import”
• “Round-trip check. Re-import what I exported and diff”

Preconditions

1. Solution open in Designer.
2. UNS has content to export. Empty solutions produce a trivially-empty graph.

Steps

1. If the user has been modifying @Tag.X.Attributes at runtime, offer promote_retentive_attributes first. The exporter reads cold-start for determinism, so runtime-only changes do not appear in the export otherwise.
2. Pick a format: rj (default, fully implemented), jsonld (falls back to rj in 10.1.5 with a warning, native in 10.1.6).
3. Call:

   export_graph_model(format="rj", output_name="my-uns")

4. Relay the outputPath back. It is the file the user shares or uploads.
5. If any warnings[] entry mentions a format fallback, surface it so the user is not surprised their .jsonld request came back as .rj.

Gotchas to surface

• No SPARQL push in 10.1.5. The file lands in the Exchange folder. Upload to GraphDB / Stardog / Fuseki is the user's job.
• Round-trip is faithful at the UserType level, not the OWL axiom level. Restrictions and anonymous class unions do not survive. Set expectations before the diff.
• FX Solution Localization strings are NOT consulted at export. Alt-language rdfs:label variants live in the Attributes column keyed rdfs:label@pt and similar.

See also

• Export your UNS to RDF/OWL/GraphDB. Human walkthrough.
• Skill Import Industrial Ontology. Reverse direction.