scientific-agent-skills hit GitHub trending at #3 today with 670+ stars in a single day. K-Dense AI bundles 135 ready-to-use Agent Skills covering scientific research workflows — and they plug directly into Claude Code, Cursor, Codex, and any agentic coding tool that reads `SKILL.md` files.
## What’s inside
The repo covers cancer genomics, drug-target binding prediction, molecular dynamics, RNA velocity analysis, geospatial science, and time-series forecasting. There are dedicated skills for scientific ML resource discovery via Hugging Science, integration with 78+ scientific databases, and structured scientific writing. Each skill is a single Markdown file that an agent reads and follows — no SDK, no Python package to install in advance.
## Why it matters
Most agent skill libraries today are toy-level: summarize a PDF, fetch a URL, format a table. K-Dense is the first comprehensive scientific skill bundle. Drop these into a Claude Code or Cursor session and your AI assistant becomes a research collaborator that can query PubChem, run RDKit chemistry computations, and chain analysis steps across biology, chemistry, and medicine.
## License
Repository under MIT. Individual skills carry their own license metadata in each `SKILL.md` — most are free for commercial and noncommercial use, but check per-skill before shipping production code.

Leave a comment