Skip to content

Run Your First Skill

Beginner ~15 min

No installation. No terminal. No cost. Everything runs in your browser via Google Colab. All you need is a Google account.

:material-open-in-new: Launch in Google Colab

1. Install ClawBio

Open a new Google Colab notebook and run:

!git clone https://github.com/ClawBio/ClawBio.git
%cd ClawBio
!pip install -q -r requirements.txt

That is it. ClawBio is ready to use.

2. Run the PharmGx Demo

Every ClawBio skill ships with synthetic demo data. Run the PharmGx Reporter — it generates a pharmacogenomics report from a sample 23andMe-format file:

!python3 skills/pharmgx-reporter/pharmgx_reporter.py \
  --input skills/pharmgx-reporter/demo_patient.txt \
  --output /tmp/pharmgx-demo

You should see output like:

✓ Loaded 47 variants from demo_patient.txt
✓ Matched 12 actionable pharmacogenes
✓ Report written to /tmp/pharmgx-demo/report.md
✓ Summary written to /tmp/pharmgx-demo/summary.json

3. Inspect the Output

!cat /tmp/pharmgx-demo/report.md

The report includes:

  • Patient summary — matched pharmacogenes and star alleles
  • Drug-gene interactions — CPIC guideline-backed recommendations
  • Safety warnings — flagged high-risk variants (e.g. DPYD*2A for fluorouracil toxicity)
  • Research-use disclaimer — ClawBio is not a medical device

View the structured JSON:

import json
with open("/tmp/pharmgx-demo/summary.json") as f:
    data = json.load(f)
print(json.dumps(data, indent=2))

4. Try Another Skill

Run any other skill with --demo:

# Polygenic risk scores
!python3 skills/gwas-prs/gwas_prs.py --demo --output /tmp/prs-demo

# Single-cell RNA-seq pipeline
!python3 skills/scrna-orchestrator/scrna_orchestrator.py --demo --output /tmp/scrna-demo

# Health equity scoring
!python3 skills/equity-scorer/equity_scorer.py --demo --output /tmp/equity-demo

# GWAS variant lookup across 9 databases
!python3 skills/gwas-lookup/gwas_lookup.py --demo --output /tmp/gwas-demo

5. Understand the Flow

Every ClawBio skill follows the same pattern:

  1. SKILL.md defines the contract — inputs, outputs, domain decisions, safety rules
  2. Python script implements the analysis — reads inputs, runs the pipeline, writes outputs
  3. Demo data provides synthetic test data — no real patient data needed
  4. An AI agent (optional) dispatches the skill, passes parameters, and explains the results

The agent reads the SKILL.md to understand what the skill does and how to call it. The agent never overrides domain decisions or safety rules defined in the SKILL.md.

Next: Build a Skill — create your own skill from scratch.