SetGo: Open-Source Toolkit for Scientific AI Dataset Metadata Readiness

SetGo is an open-source Python toolkit designed to evaluate and enhance the metadata readiness of scientific AI datasets across six key areas: FAIR compliance, licensing, provenance, governance, reproducibility, and catalog readiness. It was created to fill the void left by the Readiness Engine for Data Integration (REDI), which focuses solely on computational readiness. SetGo assesses the completeness of metadata, governance, and adherence to standards prior to publication or archiving. Unlike existing FAIR assessors that only analyze published repository records, SetGo addresses all six dimensions simultaneously. When applied to four scientific corpora, it identified shortcomings that standard tools overlook; for instance, ERA5 climate data achieved a mere 4% compliance with ACDD 1.3. The toolkit aims to make datasets ready for discovery and suitable for agent consumption, particularly for AI model training and data sharing.