ShareSift

ShareSift is a machine learning-enhanced tool designed to identify and rank files on SMB shares that are likely to contain credentials or secrets. Utilizing a two-stage classifier pipeline, it combines a LightGBM path classifier and a Qwen3 1.7B LoRA content classifier to improve recall of sensitive information significantly over its predecessor, Snaffler. Notable features include adjustable classification policies to balance false positives and recall rates, allowing users to optimize for specific operational needs.