scientific-agent-skills
Turn any AI agent into an AI Scientist. The #1 Agent Skills library for science, used by 160,000+ scientists worldwide. 140 ready-to-use skills plus 100+ scientific databases covering biology, chemistry, medicine, and drug discovery. Compatible with Cursor, Claude Code, Codex, Pi, Antigravity, and the open Agent Skills standard.
A collection of 135 pre-packaged skill instructions that teach your AI coding assistant how to use over 100 scientific databases and 70+ research Python packages, enabling reliable multi-step workflows in biology, chemistry, medical imaging, and more.
This repository is a collection of 135 ready-to-use "skills" — pre-packaged instructions and examples — that you can add to an AI coding assistant to turn it into a scientific research helper. Instead of the AI having to figure out how to use complex scientific tools from scratch, these skills give it curated documentation and examples for over 100 scientific databases and 70+ Python packages commonly used in research.
The skills cover a wide range of scientific domains: biology and genetics analysis, drug discovery and chemistry, medical imaging, clinical research, machine learning, materials science, physics, geospatial data, and more. For example, the skills give your AI agent direct access to databases like PubChem, UniProt, and ClinicalTrials.gov, along with guidance on how to use specialized tools for protein structure analysis, molecular simulation, and scientific writing.
You would use this if you are a researcher, scientist, or anyone doing technical analysis who wants an AI assistant that reliably handles complex multi-step workflows — like processing genomics data, running drug screening analyses, or writing grant proposals. The skills work with AI coding tools including Claude Code and Cursor. The companion project K-Dense BYOK (bring your own API keys) is a desktop application built on top of these skills that provides a full research workspace.
The README notes this repository is truncated, so the full README is longer than what was provided. The project is written in Python and released under an MIT license.
Where it fits
- Add biology or drug discovery skills to Claude Code so it can query PubChem or UniProt databases automatically
- Run a multi-step genomics analysis workflow using an AI assistant that already knows the right Python packages
- Enable your AI coding agent to search ClinicalTrials.gov and summarize trial data for a specific drug
- Use the companion K-Dense BYOK desktop app as a full scientific research workspace without configuring individual skills manually