A case study in temporal data leakage: deliberately builds a 99 %-accurate but useless leaky model, then rebuilds it as a leakage-safe, time-aware ML pipeline with honest cross-validation and a Streamlit deployment.

Point-in-time tool-call leakage scoring for LLM agents, as an Inspect extension. Measures whether an agent uses information it could not have had at the time.

Pre-training feature leakage auditor for tabular ML datasets. Checks column names, target correlation, categorical proxies, future timestamps, ID columns, and train/test distribution shift before any model is trained.