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

L1 document ingestion — a document to the L1 source knowledge graph

Process_L1DocumentIngestion · advisory · 12 step(s)

folio-assistant — L1 document ingestion — a document to the L1 source 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/.

CARRIES THE BODY document-ingestion.bpmn HAD until placement PR6 (bean apcg; owner ruling 3, 2026-09-30: “only basic doc ingestion high level workflow in cat-harness … more detailed doc ingestion/cataloguing methodologies in folio-asst-core”). Every activity id is unchanged, so a reference to Task_Detect or CallActivity_Extract means what it meant.

IT REFINES THE HARNESS FLOW BY CALLING IT. CallActivity_Basic runs the harness’s Process_Ingestion first — accept the upload, put its metadata into the KG, catalogue the asset in library/ — and this process then reads the document into a complete L1 entry. That is the inversion the harness skill library-ingestion describes: the call points DOWN, from core to the harness, and nothing in the harness names this process. The theme branch’s call into the harness’s ingest-theme.bpmn is downward for the same reason; the four ingest-* phases sit beside this file. The method is the l1-document-ingestion skill.

BPMN diagram: L1 document ingestion — a document to the L1 source knowledge graph

How it connects

Called by: no call activity names this process
Calls: Ingestion subprocess — build the L1 knowledge graph, Ingestion subprocess — derive content from the assets, Ingestion subprocess — extract structure, Basic ingestion — an upload to an asset catalogued in library/, Ingestion subprocess — ingest a theme, Ingestion subprocess — the L1 completeness gate
Presented on: Document ingestion — The pipeline
Skill: l1-document-ingestion

Lanes — who acts

lane	role	what it does here
Ingestion Engine (agent, runs unattended)	`ingestion-agent`	Runs the whole pipeline from CallActivity_Basic through Task_Promote, and then Task_SummaryQueue, where it drains a few prose blocks of the advisory summary queue — including the theme branch, which stays in THIS lane rather than dipping into Lane_2: deciding a document is a theme source is the engine asking its author a question, not a write to the shared work plan. It does not stop at a gap: Gateway_Complete’s ‘gap’ branch leaves this lane only long enough for Lane_2 to record the bean, then Flow_i8 returns control here at CallActivity_Derive rather than ending the run — so an incomplete L1 build is this lane’s problem to keep working until Gateway_Complete says otherwise.
Work plan — beans (shared by humans and agents)	`work-plan`	A one-task detour from the Ingestion Engine’s own loop: it exists as its own lane rather than a step inside Lane_1 specifically so that recording an incomplete L1 build is visible as a write to the SHARED plan — a bean anyone can see and pick up — rather than a private retry the engine keeps to itself.
Corpus — L1 source knowledge graph	`corpus`	The terminal state the whole diagram exists to reach: once Task_Promote lands material in library//, this lane is where that fact is recorded as available, and it is the boundary this diagram stops at — everything that depends on the source afterward, such as authoring-a-document.bpmn's citations, starts from here rather than being drawn into this process.

Steps

Every one of the 12 step(s) is documented.

step	lane	skill / sub-process	what it does
Basic ingestion (harness) `CallActivity_Basic`	Ingestion Engine (agent, runs unattended)	calls Basic ingestion — an upload to an asset catalogued in library/ `library-ingestion`	The harness’s `document-ingestion.bpmn`: accept the upload by a declared route, extract its metadata into the KG, and catalogue the asset in library/ — placed if materialized, recorded as referenced if not. Everything below refines that result for a document. Called, not copied: this was Lane_0 and Task_Place of this diagram until placement PR6 moved them down as the basic flow.
Detect media type and mint a doc id `Task_Detect`	Ingestion Engine (agent, runs unattended)	`document-intake`	Sniff the media type rather than trusting the extension, and mint the doc id from the page-1 arXiv stamp where there is one, else from the filename.
Establish the licence from the upload alone `Task_Licence`	Ingestion Engine (agent, runs unattended)	`document-intake`	EARLY, by the owner’s ruling on bean 7bg9 (2026-09-20): before any derivation, because library/ is holds: content and anything derived first is committed, so refusing it later is a deletion nobody may take unasked. This step sees only the upload itself — the licence recorded in its intake.json (the same stated / unknown record a library manifest carries) and whether a LICENSE file sits beside it. It never infers a licence from extracted text. The verdict is stated, unknown, or UNDETERMINED, and undetermined is never reported as cleared: the pipeline proceeds and says so, because a licence that only extraction can reveal is a later check’s to find.
Extract structure `CallActivity_Extract`	Ingestion Engine (agent, runs unattended)	calls Ingestion subprocess — extract structure `document-intake`	See ingest-extract-structure.bpmn.
Derive content from the assets `CallActivity_Derive`	Ingestion Engine (agent, runs unattended)	calls Ingestion subprocess — derive content from the assets `document-intake`	See ingest-derive-content.bpmn. Most of this subprocess is not implemented yet.
Ingest the theme `CallActivity_IngestTheme`	Ingestion Engine (agent, runs unattended)	calls Ingestion subprocess — ingest a theme `theme-art-intake`	See ingest-theme.bpmn. The subprocess derives one Theme node with kind sticky\|webpage\|publication, and refuses a source it cannot complete rather than emitting a partial theme.
Build the L1 knowledge graph `CallActivity_BuildKg`	Ingestion Engine (agent, runs unattended)	calls Ingestion subprocess — build the L1 knowledge graph `document-intake`	See ingest-build-l1-kg.bpmn.
L1 completeness gate `CallActivity_Gate`	Ingestion Engine (agent, runs unattended)	calls Ingestion subprocess — the L1 completeness gate `l1-document-ingestion`	See ingest-l1-completeness-gate.bpmn. Skill-backed by `l1-document-ingestion` (2026-09-30, bean 7bg9; the section moved there from the harness’s `library-ingestion` in placement PR6): its section “What a complete L1 entry holds” names `check:l1-complete` as the gate and states its three results. Until then this step named `paper-relevance-triage`, which never existed, and was left uncovered rather than bound to a guess. The gate contains one adjudication (Task_FlagDrift: `real`, `spurious` or `source-wrong`), and its own Task_Verdict records the answer as a reason against the verdict before returning. So both answers reach this step, and nothing here branches on them; Gateway_Complete branches on the gate’s completeness verdict instead.
Record the gap as a bean `Task_OpenBean`	Work plan — beans (shared by humans and agents)	`todo-manager`	A missing derived artefact is a tracked gap, not a silent omission. The document stays in uploads/ until the gap closes.
Move into library// `Task_Promote`	Ingestion Engine (agent, runs unattended)	`document-intake`	The folder name IS the bibliography citation key, so a citation and a directory are the same string.
Prose blocks enter the summary queue `Task_SummaryQueue`	Ingestion Engine (agent, runs unattended)	`l1-document-ingestion`	Owner, 2026-09-24: “Make as QA sidecar as part of general doc ingestion to slowly drain.” Nothing is written to ENQUEUE a block: the queue is derived (every prose block in every declared library, minus those whose summaries.json record is a current draft or confirmation), so a promoted entry is in it the moment its blocks are. A re-ingested document whose text changed re-enters it on its own, because the record’s source_hash no longer matches. What an agent doing ingestion work does here is DRAIN a few: `bun run cat summaries:next -- --n K` hands it the next K blocks with their text, it writes a short summary of each in its own words, and `bun run cat summaries:record` writes them into library//summaries.json as drafts naming the agent and its model. The block itself stays verbatim and `ingested`. ADVISORY, never a gate: this step does not hold up Task_Citeable, and check:l1-complete reports the backlog (`block-summaries`) without failing on it. Confirming or rejecting a draft is a person's act, in `bun run cat narratives`.
Available to cite as an L1 source `Task_Citeable`	Corpus — L1 source knowledge graph	`document-intake`	Every knowledge-graph reference to this source now resolves through library/. Authoring and review consume it from here – see authoring-a-document.bpmn.

Decisions

Every one of the 2 decision(s) is documented.

decision	what decides it	branches
A theme source? (author’s judgement) `Gateway_ThemeSource`	Wired 2026-09-23 on the owner’s ruling for bean `j66n`, choosing “human/agentic judgement at the gateway” over a declared predicate. THE PREDICATE IS DELIBERATELY ABSENT. The alternative on offer was a testable rule — a captured web deployment, or a document that STATES palette and typography rules — and it was refused for the reason the owner had already given when withdrawing `xffc`/`d3yq`: “no formal role/theme mapping per se. that is authoring (human/agentic) decision/judgement.” So this gateway is answered by whoever is ingesting, not computed. An arbitrary branded PDF is not a theme source because a rule says so; it is not one because the author says it is not. WHY IT IS A GATEWAY AND NOT A FILTER INSIDE THE SUBPROCESS. `ingest-theme.bpmn` starts at “Theme source in hand” and can refuse as incomplete. Routing every document into it would make “this is not a theme” and “this theme is malformed” the same refusal, and the second is a defect while the first is the normal case.	yes → Ingest the theme no → Build the L1 knowledge graph
L1 complete? `Gateway_Complete`	The verdict of the L1 completeness gate. `gap` records the gap as a bean; `complete` promotes the source into library//.	gap → Record the gap as a bean complete → Move into library//