MedAdapt Content Server

A specialized Model Context Protocol (MCP) server for Claude Desktop that enhances AI-assisted medical learning by fetching and processing educational resources from PubMed, NCBI Bookshelf, and user-provided documents.

Overview

The MedAdapt Content Server integrates with Claude Desktop to provide tools for searching, retrieving, and analyzing medical education content. It serves as a bridge between Claude and medical knowledge sources, allowing for enhanced AI-assisted learning experiences.

Quick Start

# Clone the repository
git clone https://github.com/ryoureddy/medadapt-content-server.git
cd medadapt-content-server

# Install dependencies
pip install -r requirements.txt

# Run the server
python content_server.py

Features

• Content Search: Search for medical educational content across multiple sources
• Resource Retrieval: Fetch complete articles, book chapters, and user documents
• Topic Overviews: Generate comprehensive overviews of medical topics
• Learning Resources: Suggest appropriate learning resources based on topic and student level
• Learning Plans: Create structured learning plans with objectives and resources
• Content Analysis: Extract key points, methodologies, and findings from medical resources
• User Content: Import and analyze user-provided documents

Installation

Standard Installation

1. Clone the repository:

git clone https://github.com/ryoureddy/medadapt-content-server.git
cd medadapt-content-server

2. Create a virtual environment (optional but recommended):

python -m venv .venv
source .venv/bin/activate  # On Windows, use: .venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Configure (optional):
   • Get an NCBI API key for improved rate limits: https://ncbiinsights.ncbi.nlm.nih.gov/2017/11/02/new-api-keys-for-the-e-utilities/
   • Create a .env file based on .env.example

Usage

Running the Server

python content_server.py

Integration with Claude Desktop

1. Open Claude Desktop
2. Go to Settings → Model Context Protocol → Add Server
3. Configure with the following JSON in your claude_desktop_config.json file located in:
   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "medadapt": {
      "command": "/path/to/python",
      "args": [
        "/path/to/medadapt-content-server/content_server.py"
      ],
      "env": {
        "DB_PATH": "/path/to/medadapt-content-server/medadapt_content.db"
      }
    }
  }
}

Replace /path/to/python with your actual Python path (e.g., /opt/anaconda3/bin/python or C:\Python311\python.exe). Replace /path/to/medadapt-content-server/ with the absolute path to your cloned repository.

Important: The DB_PATH environment variable ensures the database file is created and accessed with an absolute path, preventing common file access errors.

Populating Initial Topic Mappings

python populate_topics.py

Testing

Run tests to verify everything is working:

python test_server.py

Example Usage with Claude

Scenario 1: Learning About a Medical Topic

User prompt to Claude:

I'd like to learn about the cardiac cycle. Can you provide a big picture overview and help me understand the key concepts?

Scenario 2: Finding Specific Resources

User prompt to Claude:

I need to find recent research articles about COVID-19 treatment options. Can you help me find relevant resources?

Scenario 3: Creating a Learning Plan

User prompt to Claude:

I'm a second-year medical student studying neurology. Can you create a learning plan for understanding stroke pathophysiology?

Available Tools

The server provides the following tools to Claude:

• search_medical_content: Search for medical content with filters
• get_resource_content: Retrieve complete content for a specific resource
• get_topic_overview: Generate comprehensive overview of a medical topic
• suggest_learning_resources: Get personalized resource recommendations
• import_user_document: Upload user-provided learning materials
• generate_learning_plan: Create structured learning plan with objectives
• extract_article_key_points: Extract key findings from medical articles

Troubleshooting

Common Issues and Solutions

1. Database Connection Error

   • Symptom: sqlite3.OperationalError: unable to open database file
   • Solution: Make sure the DB_PATH environment variable is set correctly in your Claude Desktop configuration, pointing to an absolute path where the application has write permissions.

2. File Path Error

   • Symptom: No such file or directory errors
   • Solution: Ensure all paths in the Claude Desktop configuration are absolute paths without extra quotes or escape characters.

3. API Rate Limiting

   • Symptom: Slow or failed responses from PubMed or NCBI Bookshelf
   • Solution: Get an NCBI API key and add it to your .env file

4. Claude Desktop Connection

   • Symptom: Claude cannot connect to the MCP server
   • Solution: Verify the server is running in a terminal window and properly configured in Claude Desktop

Project Structure

medadapt-content-server/
│
├── content_server.py     # Main MCP server implementation
├── database.py          # SQLite database interface
├── pubmed_utils.py      # PubMed API utilities
├── bookshelf_utils.py   # NCBI Bookshelf utilities
├── populate_topics.py   # Script to populate initial topic data
├── test_server.py       # Test script
├── requirements.txt     # Python dependencies
├── .env.example         # Example environment variables
└── README.md            # Documentation

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• NCBI for providing access to PubMed and Bookshelf APIs
• Anthropic for Claude and the MCP integration capability