Story

I built a skill for OpenClaw (open source AI agent framework) that lets agents analyze raw genetic data from 23andMe, AncestryDNA, and similar services.

The motivation was privacy. Most genetic analysis tools require uploading your data to a server. This runs entirely locally with no network requests. Your DNA never leaves your machine.

It analyzes about 800 markers covering health risks (APOE, MTHFR, Factor V Leiden), drug metabolism (CYP450 enzymes, warfarin sensitivity), and ancestry (Y-DNA/mtDNA haplogroups for all major global lineages). Output is both human-readable reports and structured JSON that agents can use to help users understand their results. It also includes popular ancient DNA databases.

Built on PLINK and standard Python genomics tooling. Marker annotations come from ClinVar, PharmGKB, and the GWAS Catalog with citations to primary literature.

Not a medical device, obviously. But useful for exploring your own data without handing it to a third party.