Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Hugging Science Resource Finder

Discovers and wires up curated Hugging Face datasets, models, and Spaces across 17 scientific domains, from protein design to climate modeling.

Data & AnalyticsIntermediate★ 33,030⑂ 3,248AI score 8/10Last updated: Aug 9, 2026

What it does

This skill plugs Claude into the Hugging Science catalog (huggingscience.co) plus the hugging-science Hugging Face organization, then runs a disciplined five-step loop:

  1. Map your task to one or more of 17 topic slugs (biology, chemistry, genomics, materials-science, astronomy, climate, scientific-reasoning, …)
  2. Fetch catalog content via the bundled scripts/fetch_catalog.py or the raw llms.txt / topics/<slug>.md files
  3. Weigh 2–3 candidates on scale fit, license, modality (DNA vs. protein vs. SMILES vs. crystal structure), and recency
  4. Write working code with datasets, transformers, the HF Inference API, or gradio_client
  5. Cite the author-written methodology blog behind the chosen resource

It bundles focused reference files for datasets, models, Spaces, topic slugs, and flagship picks, and it encodes real safety habits: stream anything over a few GB, load HF_TOKEN from .env instead of hard-coding, and always ask before enabling trust_remote_code=True.

Who it's for

  • ML researchers and grad students working in biology, chemistry, materials, genomics, astronomy, or climate
  • Anyone tired of keyword-searching the Hub for "is there an open dataset/model for X?"
  • Engineers reproducing scientific ML papers or standing up benchmark evaluations
  • Not for generic ML work (recsys, chatbot RAG, cats-vs-dogs vision) — the skill explicitly defers there

Examples

  • Binder design: "design a binder for this target protein" → the skill locates the BoltzGen Space and generates a gradio_client call instead of a heavy local setup.
  • Right-sized model: "a protein language model I can run on a laptop" → it recommends ESM2 35M/650M over Evo-2 40B and writes the transformers fine-tuning script.
  • Materials exploration: fetch_catalog.py topic materials-science --filter models to list crystal-structure models, inspect the schema, then load the dataset with streaming=True.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/hugging-science folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/hugging-science/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/K-Dense-AI/scientific-agent-skills.git && mkdir -p ~/.claude/skills && cp -r scientific-agent-skills/skills/hugging-science ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and move to a working directory (e.g. cd ~/Downloads).
  2. Clone the repository: git clone https://github.com/K-Dense-AI/scientific-agent-skills.git
  3. Create the skills folder if needed: mkdir -p ~/.claude/skills
  4. Copy just this skill: cp -r scientific-agent-skills/skills/hugging-science ~/.claude/skills/
  5. Install the Python dependencies: pip install datasets transformers huggingface_hub gradio_client python-dotenv requests
  6. (Optional) For gated models and clinical datasets, create a .env file in your project with HF_TOKEN=hf_... and add .env to .gitignore.
  7. Restart Claude Code and ask something like "find an open genomics dataset I can fine-tune on" — the skill activates automatically.
  8. When the skill asks permission to set trust_remote_code=True, review the named model repo before approving, since that flag runs code from the repo on your machine.
View source on GitHub ↗License: MIT