Review gates

New Shape: which competency question requires it? · established definitions, data dictionaries, identifiers, stewardship, and user workflows reviewed? · if it departs from a mature domain model, what measurable CQ benefit outweighs interoperability and maintenance costs? · semantic contract written (identity test + counterexamples)? · Thing vs Assertion vs Collection justified? · fields described and constrained? · naming contract if hierarchical? · owning repo unambiguous?

New relationship: can a direct shaped wref answer the competency question completely? (If yes, stop.) · does the link instead have independent identity, lifecycle, attributes, or more than two participants? · only if the relationship itself is attributed or contestable: predicate declaration written (direction, member patterns, subject-name recipe)? · structural and concurrent cardinality? · epistemic class chosen? · qualifiers on the Assertion, not endpoints? · anti-inferences listed?

Direct wref binding: does the relationship freeze an exact evidence/release/schema/input version, or follow an enduring target identity? · current field behavior represented honestly as version-bound? · identity-following requirement declared as a platform dependency rather than simulated with holder revisions?

New source stream: logical Source and stream identity stable? · complete semantic projection declared? · canonicalization policy and source-grain sort key versioned? · transport-permutation and semantic-change tests pass? · canonical artifact durable and hash-verified? · raw transport excluded from knowledge and covered by a 30-day storage lifecycle? · unchanged artifact and unchanged-record no-op behavior proven? · grounded record evidence pins exact artifact versions?

Traversability (per canonical type): FROM it to context · TO it from context · WHY a direct relationship exists (the owning Thing’s grounded evidence) or why an attributed relationship claim exists (evidence on the Assertion) · a conclusion BACK to source evidence. Any “no” fails review.

New repo: is it an independently meaningful knowledge product? · do ownership, visibility, or governance require the boundary? · do release/correction lifecycle and cadence support it? · do the benefits exceed cross-repo authorization, discovery, release, and dependency costs? · which layer(s)? · reconciliation done — internal and external providers? · writer authority defined? · naming follows the repo’s NamingContracts? · Readme/Agents written?

New writer: writes only one place? · may/may-not-create explicit? · deterministic idempotent naming? · provenance on every write? · quarantine policy for the unresolved?

Changing a published contract: which downstream dependents are affected, and how do they migrate or re-evaluate?

Ontology release: fresh repository or existing history? · first release usable? · no synthesized repository-local versions? · required component Shapes installed before dependent records? · implementation work and future semantic phases explicitly separated?

Semantic closure: any unresolved semantic decision still blocks the selected release? · every accepted answer moved into its durable home? · routine choices kept out of OntologyDecision? · cross-domain principle candidates filed for review rather than silently added to local or host-specific doctrine?

New binomial proposition: precisely statable? · negation semantics defined? · temporal qualification policy? · would two rational, informed parties ever disagree? (If not, it’s a fact, not a proposition.)