notemd-knowledge-graphlisted

# notemd Knowledge Graph & Argument Maps (read-only reference) notemd has two graph-like features. **Neither is a file you author.** This skill exists so an agent understands what they are, how to influence them, and — critically — what not to fabricate. There is no `.canvas` / JSON Canvas equivalent in notemd; do not create one. ## The knowledge graph is derived, not authored notemd's graph view is generated from the notes themselves. You cannot write nodes or edges directly. Edges come from three sources: | Edge kind | Comes from | |---|---| | `articleLink` | Wikilinks `[[Name]]` between notes (see `notemd-markdown`) | | `hierarchy` | Folder-note layout — `Foo.md` beside a `Foo/` folder parents its contents | | `citation` | Citation links to indexed sources (`notemd://cite/…`, created in-app) | **To shape the graph, change the notes:** add or remove `[[wikilinks]]`, and organize folder notes. That is the only lever. Do not attempt to emit graph data structures. ## Argument maps are read-only AI output An **argument map** is not authored by hand or by an agent. notemd generates it in-app by running an extraction model over an indexed source document (typically a PDF). Its structure (`ArgumentMapDocument`): - **Nodes** with kinds: `central_claim`, `sub_claim`, `evidence`, `assumption`, `counterargument`, `rebuttal`, `limitation`, `scope_condition`, `cited_warrant`. - **Edges** with kinds: `supports`, `qualifies`, `rebuts`, `assumes`, `evidences`. - Each node carries a