Gen3 MCP Client

A Python client application that leverages the Model Context Protocol (MCP) to interact with Gen3 data services. This client enables AI-powered analysis of research studies and data using either Anthropic's Claude or Ollama models.

Features

• Connects to Gen3 data services through MCP
• Supports multiple LLM backends:
  • Anthropic Claude (claude-3-5-sonnet)
  • Ollama (qwen3:14b)
• Environment variable configuration
• Asynchronous operation
• Flexible MCP server configuration

Prerequisites

• Python 3.12 or higher
• uv package manager
• Access to either Anthropic API or Ollama

Installation

1. Clone the repository:

git clone [repository-url]
cd gen3-mcp-client

2. Install dependencies using uv:

uv sync

Configuration

1. Create a .env file in the project root with your credentials:

ANTHROPIC_API_KEY=your_api_key_here  # If using Claude

2. Configure MCP servers in mcp_servers/mds_mcp.json:

Running locally

{
    "mcpServers": {
        "gen3": {
            "command": "uv",
            "args": [
                "--directory",
                "/path/to/gen3-mcp-server",
                "run",
                "gen3.py"
            ]
        }
    }
}

Using dockerized MCP server

{
    "mcpServers": {
        "gen3": {
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "gen3-mcp-server"
            ]
        }
    }
}

Usage

Run the client:

uv run client.py

The client will:

1. Connect to the configured Gen3 MCP server
2. Initialize the specified LLM (Claude or Ollama)
3. Process queries about research studies and data
4. Output analysis results

Development

To switch between LLM backends, modify the LLM initialization in client.py:

For Claude:

llm = ChatAnthropic(model="claude-3-5-sonnet-20240620")

For Ollama:

llm = ChatOllama(model="qwen3:14b")

License

[Add your license information here]