Shap

If you need to explain why your ML model made a prediction, this gets you SHAP values and visualizations fast. Works with XGBoost, LightGBM, neural nets, basically anything. The skill walks you through picking the right explainer (TreeExplainer for tree models, DeepExplainer for neural nets, KernelExplainer for black boxes), computing feature attributions, and generating the standard plots: waterfall for individual predictions, beeswarm for global importance, scatter for relationships. The workflow sections are solid, especially the debugging and fairness analysis bits that show you how to catch data leakage or check for bias across demographic groups. Good reference if you're tired of Googling which SHAP plot does what.

Install to Claude Code

npx -y skills add k-dense-ai/scientific-agent-skills --skill shap --agent claude-code