If you're dealing with genomic variant data at scale, this wraps TileDB's sparse array technology to make VCF/BCF storage and queries actually manageable. The big win is incremental sample addition without expensive merging operations, which matters when your cohort keeps growing. Built on 1-based coordinates like VCF standard, works with cloud storage out of the box. One gotcha: it only handles single-sample VCFs and needs index files, so you'll need to split multi-sample files first. Best for population genomics workflows where you're constantly querying specific regions across hundreds or thousands of samples, not for quick one-off analyses on a handful of files.
npx skills add https://github.com/k-dense-ai/scientific-agent-skills --skill tiledbvcf