MedData API

Agent-friendly medical MCP server for AI agents that need structured, clinically accurate medical data.

What it does

Exposes a Model Context Protocol (MCP) server over stdio with 5 tools:

| Tool | Description | |------|-------------| | get_medical_concept | Physiology, anatomy, pharmacology, immunology concepts with mechanisms and clinical relevance | | get_drug_info | Drug mechanism, indications, contraindications, dosing, side effects, monitoring | | get_drug_interactions | Pairwise drug interaction check with severity, mechanism, and management | | get_icd11_code | ICD-11 diagnostic codes with inclusion/exclusion terms | | get_differential_diagnosis | Ranked differential diagnosis from a list of symptoms |

Prerequisites

• Node.js ≥ 20
• npm

Setup

npm install
npm run build
npm run seed        # populates meddata.db with initial data
npm start           # starts the MCP server (stdio mode)

Or in one step:

npm run setup && npm start

The SQLite database (meddata.db) is created in the project root on first seed.

Connect to Claude Desktop

Add the following to your Claude Desktop config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "meddata": {
      "command": "node",
      "args": ["C:/Users/bruno/meddata-api/dist/index.js"]
    }
  }
}

Restart Claude Desktop after saving. You should see "meddata" listed under MCP servers.

Tool reference

get_medical_concept

// Fetch a concept
{ "name": "action potential" }
{ "name": "RAAS", "category": "physiology" }

// Browse
{ "action": "list_categories" }
{ "action": "list_concepts", "category": "anatomy" }

Fields returned: name, category, summary, mechanism, clinical_relevance, related_concepts_list

get_drug_info

{ "name": "metformin" }
{ "name": "ACE inhibitor" }   // searches by drug class
{ "action": "list_classes" }

Fields returned: name, generic_name, drug_class, mechanism, indications, contraindications, side_effects, dosing, monitoring

get_drug_interactions

{ "drugs": ["warfarin", "aspirin"] }
{ "drugs": ["warfarin", "aspirin", "omeprazole"] }  // checks all 3 pairs

Fields returned per pair: drug_pair, interaction (with severity, mechanism, clinical_effect, management)
Severity levels: mild → moderate → severe → contraindicated

get_icd11_code

{ "query": "5A11" }                           // by code
{ "query": "diabetes", "search_by": "any" }  // by keyword
{ "query": "atrial fibrillation" }

Fields returned: code, title, description, inclusion_terms, exclusion_terms, category

get_differential_diagnosis

{
  "symptoms": ["dyspnoea", "pleuritic chest pain", "haemoptysis", "tachycardia"],
  "max_results": 5
}

Fields returned per diagnosis: name, matched_symptoms, match_score, distinguishing_features, key_investigations, icd11_code, prevalence

Seed data (initial)

| Category | Count | Notes | |----------|-------|-------| | Medical concepts | 16 | Physiology (8), Anatomy (8) | | Drugs | 8 | Metformin, Atorvastatin, Lisinopril, Aspirin, Warfarin, Amoxicillin, Furosemide, Omeprazole | | Drug interactions | 10 | Clinically significant pairs | | ICD-11 codes | 10 | Common conditions across specialties | | Diagnoses (differential) | 8 | PE, CAP, DVT, T2DM, HFrEF, AF, Appendicitis, Hypothyroidism |

Development

npm run dev     # TypeScript watch mode (recompiles on change)
npm run seed    # Re-run seed (uses INSERT OR IGNORE — safe to re-run)

Extending the data

All seed data lives in src/data/. Add entries to the relevant file and re-run npm run seed. The schema is in src/db/database.ts.

To add a new tool:

1. Create src/tools/your_tool.ts with the query logic
2. Register it in src/index.ts under ListToolsRequestSchema and CallToolRequestSchema
3. Rebuild: npm run build

Medical disclaimer

This server provides educational and reference information for AI agents and medical students. It is not a substitute for clinical judgment, current prescribing guidelines, or peer-reviewed literature. Always verify critical clinical data against authoritative sources (BNF, UpToDate, local formulary).