Summary

• Operation.process_json now parses a top-level dict carrying any JSON-LD reserved key (@context, @graph, @id, @type) into an rdflib.Graph, ahead of the existing generic-dict recursion. Symmetric with the bare-value auto-wrap (json_to_rdflib): a JSON scalar becomes a Literal; a JSON-LD object becomes a Graph.
• Previously every consuming op (POST, PUT, Merge, ldh-CreateItem, ldh-CreateContainer, the ldh-Add* family) re-parsed the same JSON-LD payload in its own execute_json. The runtime treated bare JSON-LD as opaque JSON to recurse into; only the ops at the boundary knew its semantics.
• Unblocks the chat-extraction use case in AutoGraph: an LLM that emits JSON-LD as its final answer (image-extracted structured metadata) now gets a real Graph result the renderer can lay out — instead of a dict of Literals that renders as raw text.
• Version bumped to 1.3.0.

Test plan

• Existing unit suite — uv run pytest tests/unit -q → 96 passed, 0 failed.
• Smoke trace — a bare JSON-LD dict (@context + @graph describing a schema:Car) handed to Operation.process_json now returns an rdflib.Graph with the expected triple count, instead of a dict.
• Once tagged and consumed by AutoGraph, end-to-end chat verification: an LLM emitting structured JSON-LD as its final answer renders as a structured per-subject view via render_graph.

🤖 Generated with Claude Code