scientific-agent-skills
K-Dense-AI/scientific-agent-skills · Python · MIT
Domain skills that let an agent do real scientific analysis.
A vertical skill pack focused on biology and data science: sequence analysis, cheminformatics, stats and reproducible research workflows an agent can run.
Score
Setup at a glance
cp -r skills/* ~/.claude/skills/ Deep dive
A skill pack that turns a general coding agent into a domain scientist. Instead of one monolithic prompt, it ships 177 curated skills across bioinformatics, cheminformatics, proteomics, clinical research, neuroscience, microscopy and machine learning — each one documenting the exact Python stack, dependency versions and validation scope for a real research task.
The skills follow the open Agent Skills standard, so they install into Claude Code, Cursor, Codex and other hosts with a single command (npx or gh skill) and run entirely on your own machine and data. Under the hood the agent chains skills into multi-step workflows: query ChEMBL or ClinicalTrials.gov, analyse structures with RDKit, dock with DiffDock, run differential expression with PyDESeq2, and write the report — with every step traceable to a source.
It is research-only by design: clinical skills produce source-bound drafts for qualified review rather than diagnoses or treatment recommendations. Versioned releases, CI tests per skill and a published arXiv paper make it reproducible enough for serious lab use, not just demos.