internscience/awesome-scientific-skills

---

The Problem

Thousands of Agent Skills are scattered across official registries and community repos. A researcher who needs to run a gene ontology enrichment, draft a grant proposal, and plot dose-response curves shouldn’t have to dig through 30,000+ entries to find the right three skills. And once found, they shouldn’t have to wire them together from scratch.

The Goal

Build a single, clone-ready repository of high-quality skills selected and organized for scientific research — spanning bioinformatics, cheminformatics, data analysis, scientific writing, literature search, and beyond.

Phase 1 (current): A curated link collection — the reading list before the library.

Phase 2: Import and reorganize selected skills into a unified directory structure.

Phase 3: Composition recipes — multi-skill workflows for real research tasks.

---

Why This Repo?

---

Skill Sources

We draw from two tiers of upstream repositories:

Official Skill Registries

Community Curations

Science-Specific

Specialized Domain Tools (Top Priority)

Academic Writing

Literature Search

Neuroscience

---

Scientific Skill Categories

Below is the target taxonomy. Skills from the sources above are mapped into these categories. Checked boxes indicate categories already well-covered upstream; unchecked indicate gaps we aim to fill.

Wet-Lab & Life Sciences

• [x] Bioinformatics & Genomics — sequence analysis, single-cell RNA-seq (Scanpy), variant annotation, phylogenetics, gene regulatory networks
• [x] Cheminformatics & Drug Discovery — molecular property prediction (RDKit), docking, ADMET, compound library search (ChEMBL)
• [x] Structural Biology — protein structure prediction, PDB retrieval, molecular visualization
• [x] Clinical & Medical Research — ClinicalTrials.gov search, ICD coding, adverse event analysis, COSMIC/TCGA queries

Computational & Data Science

• [x] Statistical Analysis — hypothesis testing, Bayesian inference, survival analysis, power calculations
• [x] Machine Learning for Science — PyTorch Lightning workflows, scikit-learn pipelines, hyperparameter tuning
• [ ] Simulation & Modeling — molecular dynamics, finite element analysis, agent-based modeling
• [ ] High-Performance Computing — job submission (SLURM/PBS), distributed training, GPU profiling

Literature & Knowledge

• [x] Literature Search & Retrieval — PubMed, arXiv, Semantic Scholar, CrossRef, DOI resolution
• [x] Scientific Writing — paper drafting, grant proposals, peer review assistance, citation management
• [ ] Knowledge Graphs — ontology navigation (Gene Ontology, MeSH), entity linking, relationship extraction

Research Operations

• [x] Data Wrangling — tabular data cleaning, format conversion, missing data imputation
• [x] Visualization — publication-quality plots (Matplotlib/Seaborn), interactive dashboards, figure layout
• [ ] Reproducibility — environment capture, notebook-to-pipeline conversion, data provenance tracking
• [x] Document Processing — PDF extraction, DOCX/PPTX generation, LaTeX compilation

Quantitative & Financial Research

• [x] Financial Data — SEC EDGAR, Alpha Vantage, OFR Hedge Fund Monitor, Treasury fiscal data
• [ ] Econometrics — time series analysis, causal inference, panel data methods

---

Quick Start

Option 1 — Browse and pick

Explore the [Skill Sources](#Skill Sources) above, find what you need, and copy individual skill folders into your project:

your-project/
└── .claude/
    └── skills/
        ├── pubmed-search/
        ├── rdkit-molecular-analysis/
        └── scientific-writer/

Option 2 — Clone the science stack (Phase 2)

# Coming soon — once we consolidate skills into this repo
git clone https://github.com/InternScience/Awesome-Scientific-Skills.git
cp -r Awesome-Scientific-Skills/skills/ your-project/.claude/skills/

Compatibility

Skills follow the open Agent Skills specification. They work with:

• Claude Code — /skills or .claude/skills/ directory
• Cursor — reads from .claude/skills/ and .codex/skills/
• Codex — .codex/skills/ directory
• Gemini CLI / Antigravity — compatible via the Agent Skills standard

---

Selection Criteria

Not every skill makes the cut. We apply these filters:

---

Contributing

Contributions are welcome. This is a curated list — quality over quantity.

To suggest a skill:

1. Fork this repository
2. Add the skill to the appropriate section in readme.md with format:

   | [skill-name](https://github.com/...) | Brief description of what it does |
   

3. Submit a Pull Request with a short note on why this skill is useful for scientific research

What we’re looking for:

• Skills that solve real problems researchers face daily
• Skills that compose well with others (e.g., a search skill whose output feeds into an analysis skill)
• Skills maintained by their authors — not abandoned after initial commit

What we don’t accept:

• Crypto/blockchain/DeFi skills (off-scope)
• Skills with no documentation
• Bulk-generated or spam skills
• Skills that require proprietary backends without free tiers

---

Security Notice

Skills can execute arbitrary code in your agent’s environment. Always review a skill’s SKILL.md and any bundled scripts before installing. The skills listed here are curated, not audited — they may be updated or modified by their original maintainers at any time.

---

Roadmap

• [x] Phase 1 — Curated link collection with source taxonomy
• [ ] Phase 2 — Community skill registry with ratings and usage metrics
• [ ] Phase 3 — Test skill in OpenClaw and Claude Code
• [ ] Phase 4 — Develop skill creator which can create skill from paper and agentic trace

---

Related Projects

Companion open-source tools from our team for skill generation and repository evaluation:

---

License

This curated list is released under MIT. Individual skills retain their own licenses as specified in their respective SKILL.md files.

---