Scientific Agent Skills (K-Dense) 🧩 Skill Free

What it does

Scientific Agent Skills, by K-Dense, is a large collection of skills that turns a general coding agent into a research assistant. The project used to be called Claude Scientific Skills. Each skill gives the agent curated documentation, examples and working patterns for a scientific Python library, database or research method. The agent can still use any package without a skill, but the skills make multi-step analyses in biology, chemistry, medicine and other fields more reliable.

What is inside

At the time of writing (September 2026) the README lists 166 skills and access to more than 100 scientific databases. The coverage includes:

• Bioinformatics and genomics, including single-cell RNA-seq and variant annotation
• Cheminformatics and drug discovery, proteomics, and protein engineering
• Clinical research, medical imaging and pathology (all framed as research-only)
• Materials science, physics and astronomy, and geospatial analysis
• Lab automation, statistics and visualisation, and scientific writing

The repository also bundles Anthropic's four document skills. These remain under Anthropic's own licence terms.

Works with

The README says the skills work with any agent that supports the Agent Skills standard and names Claude Code, Claude Cowork, Codex, Cursor, Gemini CLI and Google Antigravity. The repository is also packaged as an Agent Plugin for clients that support that format.

How to install

npx skills add K-Dense-AI/scientific-agent-skills

With GitHub CLI 2.90 or later you can install a single skill, for example:

gh skill install K-Dense-AI/scientific-agent-skills scanpy

The skills use uv to install their Python dependencies.

Maintenance and safety

The collection is MIT licensed (the vendored document skills excepted), frequently released, and described in an arXiv paper. The maintainers warn that many skills now come from community contributors and have not all been reviewed as closely. Skills can install packages, call external APIs and run code, so read each SKILL.md first. The README also suggests scanning skills before you install them and pinning a release.

Who should use it

Researchers, data scientists and lab engineers who use coding agents for analysis rather than app building. Install a topical subset rather than all 166 skills, both for safety and to keep the agent's context small.