deepchem

Quick Review

Exceptional skill for molecular machine learning and drug discovery. The description clearly articulates capabilities, making it easy for a CLI agent to determine when to invoke this skill. Task knowledge is outstanding with comprehensive coverage of data loading, featurization decision trees, model selection guides, splitting strategies, and complete workflows. The structure is excellent with clear sections, decision trees, comparison tables, and logical organization, plus well-referenced external files for deep-dive documentation. Novelty is high—molecular ML workflows require specialized domain knowledge (scaffold splitting, graph featurizers, ADMET prediction) that would consume many tokens for a general CLI agent to execute correctly. Minor deduction only because some edge cases in GNN architecture selection could be more explicit, but overall this is a production-ready, highly valuable skill that meaningfully reduces complexity and cost for drug discovery ML tasks.