Generated from folio-assistant-core/processes/library/ingest-build-l1-kg.bpmn by gen-processes-viz.ts — do not edit here. All processes

Ingestion subprocess — build the L1 knowledge graph

Process_BuildL1Kg · advisory · 5 step(s)

Build the L1 knowledge graph from what ingestion derived: Dublin Core, the manifest, the assets, their binding and their links. folio-assistant — Ingestion subprocess — build the L1 knowledge graph.

Source of truth: this file. Open it in bpmn.io, Camunda Modeler, or any other BPMN 2.0 tool. The SVG under docs/assets/img/workflows/ is generated from it by bun run cat render:bpmn — never hand-edit the SVG.

The extension on an activity names the folio-assistant skill that implements it; marks a step that reads or writes the shared work plan in beans/.

How it connects

• Called by: L1 document ingestion — a document to the L1 source knowledge graph
• Calls: none
• Presented on: Document ingestion — Build the L1 knowledge graph

Lanes — who acts

lane	role	what it does here
Ingestion Engine (agent, runs unattended)	ingestion-agent	Writes four records in a fixed order and touches the corpus graph in none of them — dublin-core as the record of truth, a manifest that references it, assets[] kept as a SIBLING of library: because that tree is regenerated and would drop anything written inside it, and finally the slug binding Task_Link depends on. Getting the order or the placement wrong here is invisible until the next sync silently drops it.
Corpus — L1 source knowledge graph	corpus	The single moment the four records Lane_0 wrote become reachable by reference rather than just present on disk: Task_Link is what a knowledge-graph reference in a paper, an L2 or an L3 artefact actually resolves through, and until it runs, dublin-core.jsonld and its siblings are files in library/ that nothing in the graph points to yet.

Steps

Every one of the 5 step(s) is documented.

step	lane	skill / sub-process	what it does
Write dublin-core.jsonld (the record of truth) Task_Dublin	Ingestion Engine (agent, runs unattended)	document-intake	One standalone Dublin Core record per folder. dcterms is already this corpus’s JSON-LD vocabulary, so this extends a live context rather than introducing one.
Write manifest.jsonld referencing it Task_Manifest	Ingestion Engine (agent, runs unattended)	document-intake	Write manifest.jsonld for the folder: its @id, @type folio:SourceDocument, what it contains and its provenance, referencing the standalone dublin-core.jsonld as the record of truth rather than copying its fields.
Record assets[] — local path or remote URL Task_Assets	Ingestion Engine (agent, runs unattended)	document-intake	kind / role / url-or-path / checksum / retrieved. A SIBLING of library:, never inside it: LibraryRef is regenerated from the tree, so an authored URL placed there is dropped on the next sync.
Bind the folder name to the bibliography slug Task_Bind	Ingestion Engine (agent, runs unattended)	document-intake	Bind the folder name to the document’s bibliography slug, so library// and its citation key are the same string: a citation resolves to a directory without a lookup table between them.
Link L1 nodes into the corpus graph Task_Link	Corpus — L1 source knowledge graph	document-intake	Anything in library/ is L1 source content. A knowledge-graph reference from any artefact – paper, L2, L3 – resolves to it through library/.