drug-interaction-mcp

An MCP server that lets AI assistants check drug interactions against a clinically sourced database of 150+ interactions.

Why?

AI assistants frequently give incorrect or incomplete drug interaction advice. When someone asks "Can I take ibuprofen with warfarin?", the answer matters — it could prevent a serious bleeding event.

This MCP server gives AI assistants (Claude, ChatGPT, etc.) access to a curated, evidence-based drug interaction database sourced from FDA drug labels, NIH DailyMed, and established medical literature. Instead of guessing, the AI can query real data.

Quick Start

Install

pip install drug-interaction-mcp

Or run directly with uvx:

uvx drug-interaction-mcp

Configure in Claude Desktop

Add to your Claude Desktop configuration file (claude_desktop_config.json):

{
  "mcpServers": {
    "drug-interactions": {
      "command": "uvx",
      "args": ["drug-interaction-mcp"]
    }
  }
}

That's it. Claude will now have access to drug interaction checking tools.

Tools

check_interaction

Check if there is a known interaction between two drugs.

check_interaction(drug_a="warfarin", drug_b="aspirin")

Returns severity, clinical description, mechanism of action, recommendation, and source.

Example response:

{
  "drug_a": "warfarin",
  "drug_b": "aspirin",
  "found": true,
  "severity": "danger",
  "description": "Significantly increased risk of major bleeding events...",
  "mechanism": "Aspirin inhibits platelet aggregation via COX-1 while warfarin inhibits vitamin K-dependent clotting factors...",
  "recommendation": "Avoid combination unless specifically directed by a physician...",
  "source": "FDA Drug Label — Warfarin"
}

check_multiple

Check all pairwise interactions among a list of drugs. Useful when reviewing a patient's full medication list.

check_multiple(drugs=["warfarin", "aspirin", "lisinopril", "metformin"])

Returns all found interactions sorted by severity (danger first), with a summary of counts.

search_drug

Fuzzy search for drug names in the database. Handles partial names, typos, and brand/generic name lookups.

search_drug(query="prozac")
search_drug(query="warfa")

get_drug_info

Get all known interactions for a specific drug.

get_drug_info(drug_name="warfarin")

Returns the drug name, total interaction count, and full details on every known interaction.

list_interactions

List all interactions in the database, optionally filtered by severity.

list_interactions(severity="danger")
list_interactions(severity="all")

Severity levels: "danger", "warning", "caution", "all".

Database Coverage

The embedded database includes 150+ clinically documented interactions covering:

• Blood thinners — warfarin, apixaban, rivaroxaban, clopidogrel + various interacting drugs
• Cardiac drugs — digoxin, amiodarone, beta blockers, calcium channel blockers, statins, nitrates
• Psychiatric drugs — SSRIs, SNRIs, MAOIs, antipsychotics, benzodiazepines, mood stabilizers
• Antibiotics — fluoroquinolones, macrolides, tetracyclines, metronidazole, rifampin, linezolid
• Diabetes drugs — metformin, insulin, sulfonylureas, SGLT2 inhibitors
• Pain medications — opioids, NSAIDs, acetaminophen, tramadol
• Supplements — St. John's Wort, ginkgo, ginseng, fish oil, iron, calcium, vitamin K/D/E, CoQ10, turmeric
• OTC drugs — antacids, PPIs, diphenhydramine, dextromethorphan

Each interaction includes:

• Drug name pairs (generic + brand names)
• Severity level (danger / warning / caution)
• Clinical description
• Mechanism of action
• Patient-facing recommendation
• Evidence source (FDA, NIH, etc.)

Data Sources

• FDA Drug Labels — Official prescribing information
• NIH DailyMed — National Library of Medicine drug information
• FDA Drug Safety Communications — Safety alerts
• NIH Office of Dietary Supplements — Supplement interaction data

Disclaimer

This tool is for informational and educational purposes only. It is not a substitute for professional medical advice, diagnosis, or treatment. Always consult a qualified healthcare provider (physician, pharmacist) before making medication decisions. The database may not include all known interactions, and individual patient factors affect interaction risk.

Development

# Clone the repository
git clone https://github.com/pillright/drug-interaction-mcp.git
cd drug-interaction-mcp

# Install in development mode
pip install -e ".[dev]"

# Run tests
pytest -v

# Run the server locally
python -m drug_interaction_mcp.server

Contributing

Contributions are welcome. To add new interactions:

1. Fork the repository
2. Add interactions to src/drug_interaction_mcp/interactions.py
3. Ensure each interaction has: drug names, severity, description, mechanism, recommendation, and a credible source
4. Add tests for new interactions
5. Open a pull request

All interaction data must be sourced from FDA drug labels, NIH, or peer-reviewed medical literature.

License

MIT