Reproducibility data and code for: A sulfur-content paradox in lithium solid electrolytes: provenance-aware machine learning with cross-database validation

Reproducibility archive supporting the study “A sulfur-content paradox in lithium solid electrolytes: provenance-aware machine learning with cross-database validation”. The archive contains processed analysis data, provenance and overlap audits, machine-learning and statistical-analysis code, cross-database validation results, GAN-ablation outputs, figure-generation scripts, environment information, and integrity checks. Synthetic data generated during GAN ablation are used only for methodological evaluation and are not treated as experimental evidence.