This is a structured clinical reasoning system for rare disease diagnosis that converts patient symptoms into HPO terms, then searches across Orphanet, OMIM, and ClinVar to generate ranked differential diagnoses. It walks through phenotype matching, gene panel prioritization with ClinGen validation, and ACMG variant interpretation with computational predictors. The workflow is opinionated: it creates a report file first and updates it progressively, insists on searching databases rather than guessing, and emphasizes multi-system involvement and treatable conditions. Useful when you're building diagnostic support tools or working with genetic counseling data. The clinical reasoning framework is solid, with practical heuristics like looking for regression patterns and occupational exposures that can have 50-year latencies.
npx skills add https://github.com/mims-harvard/tooluniverse --skill tooluniverse-rare-disease-diagnosis