This document records the authoritative document hierarchy for the substrate, the naming/containment rule that applies to all new content, and pointers to the frameworks that operationalise both.
When two documents appear to disagree, this is the order of authority:
<layer>/README.md) — the authoritative literate specification for that layer’s mechanism. spec/*.ttl, vocab/*-vocab.ttl, and shapes/*.ttl are mechanically extracted from a README’s fenced blocks and never diverge from it independently.docs/architecture/* — cross-layer reference material, kept in sync with the ADRs and layer READMEs, but not itself a source of new decisions.Substrate specifications never cite or quote applied-layer, deployment-specific, or historical design documents. A substrate document may state a generic premise and a generic law; it does not reference where that premise happened to first come up.
One rule, stated once here rather than repeated per layer or per gate:
Substrate directories (ontology/foundation/, ontology/vocabulary/, ontology/quantification/, ontology/party/, ontology/eligibility/, ontology/instrument/, ontology/behaviour/, and their spec/, vocab/, shapes/, projection/, ontology/examples/) stay domain-neutral. No class, property, named individual, fnd:utility string, or shipped example names a specific industry, product, brand, or closed-estate identifier.
Applied-layer or deployment-specific content — a concrete dimension set, a worked deployment example, a projection binding for a specific applied ontology — may be authored anywhere in this repository the user directs (for example, under WINGMAN/, which is not tracked by version control). Generating such content is not restricted; whether it is ever committed to a public branch is the user’s decision, made per instance, not gated by a standing review process. This replaces an earlier, heavier containment posture (physical repository separation, an adversarial two-role clean-room review) that this project does not require.
The authoring order for substrate content — generic premise, then non-domain examples, then mechanism prose — is fixed procedurally in ADR-A-C2, not repeated here.