Claim-Evidence-Boundary Records for Public-Source AI-Governance Review: v0.3 Reproducibility Package

Version 0.3 — 22 July 2026 This release replaces the earlier F1000Research-oriented derived-data package under the same Zenodo concept DOI. It is aligned to the substantially revised research article Claim-Evidence-Boundary Records for Public-Source AI-Governance Review: Cross-Case Evaluation and a Prospective Transfer Test. Adds a general CEB v0.3 JSON Schema and one fully populated CEB record for each of the three evaluated cases. Adds a prospective held-out C2PA standards-text transfer record that conforms to the frozen schema without an amendment. Adds the Case-1 citation-receipt bundle and reproduction outputs, the Case-2 frozen OSV/PyPI inputs and normalized clock table, and the Case-3 frozen repository-visibility corpus, codebook, and aggregates. Adds row-level author-side audit records, the bibliography audit, exact literature-search queries, retained candidate metadata, and the final inclusion ledger. Replaces the earlier conceptual figure with the corrected CEB diagram: the institutional decision is a downstream consumer, not a CEB field or an automatic result. The package supports record-level conformance, stated deterministic checks, and bounded reproduction. It does not establish legal correctness, compliance, currentness after the capture date, organizational practice, an institutional decision, or independent human-review agreement. This record contains the underlying data and extended materials for a F1000Research submission candidate on evidence instrumentation for AI-governance review. Materials include a citation-receipt JSON Schema, normalized OSV/PyPI evidence-clock rows, a frozen 100-repository evidence-visibility table, category summary data, three article figures, a claim table, source ledger, manifest, reuse notes, and a lightweight package-validation script. The materials are public-source review artifacts. They improve inspectability as an initial step before determining legal compliance, semantic truth, safety, security, quality, maintainer diligence, or institutional endorsement.