Evidence: data/review-20260914.cvXJSA/REVIEW.md, snapshot data/merged/20260914/merged-kg.tar.gz, SHA-256 0fb80d5d865d0da503d9c759286c91216b3dab5f5cc5df0a8e759cd4e233ec81 (3,353,182 nodes / 15,295,064 edges). Counts describe the immutable reviewed archive, not a rebuilt output.
19,095 anonymous NamedThing nodes include 18,374 expected IMG and 25 expected GTDB taxonomy references, leaving 696 unexpected: NCBITaxon 513, GO61, assembly60, local ingredient54, local compound3, CHEBI2, MICRO1, OBI1, gold1. Edge endpoint closure is insufficient: KGX invents these rows. stub-audit.json contains per-ID incident source/predicate evidence.
Assembly reconciliation already belongs to #1064; do not duplicate or assume GCF/GCA string substitution establishes identity. Local fallback ingredient/compound IDs whose raw labels are known should be declared by the producing transform. Ontology/taxon release gaps require source-aware resolution or explicit retained-gap evidence, not invented labels/whole-prefix exemptions.
Acceptance: reproducible per-ID diagnostic report (not just three samples per prefix), source/predicate attribution, named typed nodes for locally minted supported entities, and unresolved release/trim cases remain visible. Maintain zero orphan endpoints and narrow expected-stub policy. Do not falsely claim all 696 can be resolved from this archive alone.
Evidence:
data/review-20260914.cvXJSA/REVIEW.md, snapshotdata/merged/20260914/merged-kg.tar.gz, SHA-2560fb80d5d865d0da503d9c759286c91216b3dab5f5cc5df0a8e759cd4e233ec81(3,353,182 nodes / 15,295,064 edges). Counts describe the immutable reviewed archive, not a rebuilt output.19,095 anonymous NamedThing nodes include 18,374 expected IMG and 25 expected GTDB taxonomy references, leaving 696 unexpected: NCBITaxon 513, GO61, assembly60, local ingredient54, local compound3, CHEBI2, MICRO1, OBI1, gold1. Edge endpoint closure is insufficient: KGX invents these rows. stub-audit.json contains per-ID incident source/predicate evidence.
Assembly reconciliation already belongs to #1064; do not duplicate or assume GCF/GCA string substitution establishes identity. Local fallback ingredient/compound IDs whose raw labels are known should be declared by the producing transform. Ontology/taxon release gaps require source-aware resolution or explicit retained-gap evidence, not invented labels/whole-prefix exemptions.
Acceptance: reproducible per-ID diagnostic report (not just three samples per prefix), source/predicate attribution, named typed nodes for locally minted supported entities, and unresolved release/trim cases remain visible. Maintain zero orphan endpoints and narrow expected-stub policy. Do not falsely claim all 696 can be resolved from this archive alone.