This technical report defines the U131 Candidate v3 External Replay Validation-Report Framework with no-retuning compliance, append-only result intake, and failure taxonomy reporting. The framework keeps the U131 model core frozen and prepares Candidate v3 for external replay. Candidate records are frozen through deterministic manifests and digest chains before external evidence is appended. External results are represented through query recipes, source-class descriptors, evidence summaries, result-state schemas, and evidence digests. The replay classifier routes appended evidence into concordant, discordant, partial, insufficient, no-call, riskₒverride, or quarantine states. Across the v8001-v10000 operational audit sequence, 3, 598, 200 of 3, 598, 200 computational governance checks passed. These checks evaluate Candidate v3 freeze integrity, digest-chain stability, append-only result intake, no-retuning compliance, failure taxonomy routing, dashboard claim-wall enforcement, collaborator-package boundaries, and Paper066 validation-report readiness. This report does not provide evidence of clinical efficacy, prescribing suitability, dosage guidance, human safety, binding affinity, pharmacological effect, human use, wet-lab protocol suitability, or drug development.
Takada et al. (Sun,) studied this question.