This is a multi-biomarker integration tool for predicting whether cancer patients will respond to checkpoint inhibitors like pembrolizumab or nivolumab. It ingests tumor mutational burden, MSI status, PD-L1 expression, and resistance mutations, then outputs a 0-100 ICI Response Score with drug-specific recommendations. The workflow is thorough: it checks FDA biomarker approvals, applies cancer-specific thresholds (melanoma and NSCLC have different PD-L1 cutoffs), flags resistance genes like STK11 and JAK1/2, and distinguishes hot from cold tumors. Useful when you need to justify an immunotherapy decision with quantitative evidence rather than gut feel. The scoring tables and resistance penalty system are well documented, though you'll need actual patient biomarker data to get meaningful output.
npx skills add https://github.com/mims-harvard/tooluniverse --skill tooluniverse-immunotherapy-response-prediction