asset-pipeline-ASK_ontology-tree.source.js
This view shows one file as text; the diagram front door links it. It is generated at build time, static and bounded: nothing here runs the code, and there is no browsing beyond this file.
/* asset-pipeline-ASK_ontology-tree.source.js Source data for the Ontology diagram — Axis A. Renders the KINDS of information the definition layer structures, independent of where they live (Axis B / inheritance) or when they are produced (Axis D / runtime). Conceptual basis: docs/layer-disambiguation-note-v1.md. Rendered by the horizontal diagram-tree engine (diagrams-engine.js, window.DIAGRAMS.render) — the same top-aligned left-to-right cascade as the architecture tree. A classification (a wide branching tree, not a chain) reads best as a horizontal cascade. This is a STRUCTURAL map, not a maturity-state diagram: no `status` is asserted on any node. Do not introduce earned/held/legacy here — that is the v9-class interactive surface's job, on a different axis. Illustrative, not source truth — repo prose (README.md, docs/architecture.md) is authoritative. Bump the source-v / render-v stamp in the HTML when topology or major content changes. Node shape: { kind: 'root'|'section'|'group'|'node', label, note?, tag?, children? } */ window.TREE_ONTOLOGY = { kind: 'root', label: 'Definition Layer · Ontology', note: 'Axis A — the kinds of information the definition layer structures. The interface is a view on top of it.', children: [ { kind: 'section', label: 'Intent', children: [ { label: 'business intent', note: 'the commercial objective the asset serves' }, { label: 'creative intent', note: 'the creative framing and idea' }, ], }, { kind: 'section', label: 'Inputs', children: [ { label: 'product truth', note: 'source facts about the product' }, { label: 'visual references / invariant anchors', note: 'high-bandwidth visual guidance prose cannot encode' }, ], }, { kind: 'section', label: 'Constraints', children: [ { label: 'structural', note: 'asset types · shot grid · composition / crop' }, { label: 'aesthetic', note: 'style-guide rules · approved references acting as constraint' }, { label: 'technical', note: 'resolution · formats · naming · DAM' }, ], }, { kind: 'section', label: 'Creative discretion', note: 'the bounded, reviewable judgment exercised within the granted aperture and at selection to close it — weak discretion (judgment within standards); not intent, and not the aperture itself', children: [ { label: 'exercised within the granted aperture', note: 'the brief grants the permitted variance the executor may range over' }, { label: 'aperture = permitted variance', note: 'granted by the brief / source-of-intent package; a quantity distinct from the judgment exercised within it' }, { label: 'aimed by creative intent', note: 'intent sets the direction the aperture points; discretion is not intent itself' }, { label: 'bounded by constraints', note: 'discretion operates inside the constraint envelope' }, { label: 'shaped by references', note: 'approved references guide the permitted variance' }, { label: 'assigned to decision ownership', note: 'a decision owner is granted or withheld the latitude' }, { label: 'exercised at selection; recorded at governance', note: 'the decision makes the candidate true to intent; governance records it — selection is not governance' }, ], }, { kind: 'section', label: 'Required outputs', children: [ { label: 'required asset set / derivatives', note: 'the deliverable shape' }, { label: 'slot-specific variation', note: 'per-slot output obligations' }, ], }, { kind: 'section', label: 'Governance', children: [ { label: 'decision ownership', note: 'who owns each decision' }, { label: 'approval / override boundaries', note: 'auto-approved vs reviewed; where human judgment supersedes' }, { label: 'capture / governance rules', note: 'elevation of raw generation into a governed asset' }, { label: 'reuse / derivative / family-coherence rules', note: 'how governed outputs reuse and cohere' }, ], }, ], };