Open‑Source Skills Close the Reasoning Gap for Health‑AI Agents

AI agents built on foundation models often misapply health‑care decision frameworks even when they have the guidelines in front of them. A collection of 38 open‑source agent skills now supplies structured reasoning for 11 health domains, helping agents win 70‑86 % of head‑to‑head tests and closing the silent‑failure gap. The skills are packaged as markdown SKILL files, covering everything from genomics to claims operations.