Lightweight Artifact Verification for Cultural-Heritage Repositories: From the Modular Artifact Sheet to a Constraint-First, Evidence-Bounded Repository Model

Digital access has made cultural objects easier to see, but visibility is not the same as auditability. Object records can combine identity, chronology, findspot, provenance, attribution, transcription, and interpretation even when those elements do not have equal evidentiary support. Meanwhile, museum, archive, library, and repository sources document constrained staffing, incomplete documentation, uncertainty, fragmented information, and pressure to improve digital workflows without creating unsustainable new work. This paper develops a lightweight repository adaptation of the Modular Artifact Sheet and the later V6 evidence-bounded review process. The proposal is selective: high-interest or high-risk record components are reviewed first; existing evidence is reused; specialist judgment is reserved for questions that require it; bounded AI is used only for repeatable checks or independent second review; and the result is summarized as Green / Yellow / Red / STOP while the evidence trail remains inspectable. Established work on preservation, collection review, semantic interoperability, uncertainty, and human-centered automation informs the design without replacing it. The paper proposes a ten-to-twenty-five-record pilot to test whether the workflow improves auditability and reduces or contains professional burden enough to justify deeper adoption.