Contextual MCP Server

🚀 Contextual MCP Server

A Model Context Protocol (MCP) server that provides Retrieval Augmented Generation (RAG) capabilities using Contextual AI. This server can be integrated with various MCP clients. In this documentation, we'll demonstrate its integration with Cursor IDE and Claude Desktop.

🚀 Quick Start

Overview

This MCP server acts as a bridge between AI interfaces (such as Cursor IDE or Claude Desktop) and specialized Contextual AI agents. It implements the following:

1. Query Processing: Sends your domain - specific questions directly to a dedicated Contextual AI agent.
2. Intelligent Retrieval: Searches the knowledge base through comprehensive information.
3. Context - Aware Responses: The generated answers:

• Are rooted in the source documents.
• Include references and attributions.
• Maintain the conversation context.

Integration Process

Cursor/Claude Desktop → MCP Server → Contextual AI RAG Agent
        ↑                  ↓             ↓                         
        └──────────────────┴─────────────┴───────────────Response with references

Prerequisites

• Python 3.10 or higher.
• Cursor IDE and/or Claude Desktop.
• Contextual AI API key.
• MCP - compatible environment.

📦 Installation

1. Clone the repository:

git clone https://github.com/ContextualAI/contextual-mcp-server.git
cd contextual-mcp-server

2. Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # Use `.venv\Scripts\activate` on Windows

3. Install the dependencies:

pip install -r requirements.txt

Configuration

To modify the server settings:

1. Add or modify the following in config.py:

   # Example configuration
   API_KEY = "your_api_key"
   AGENT_ID = "your_agent_id"
   

2. Use the .env file to store the API key and AGENT_ID:

   export API_KEY=your_api_key
   export AGENT_ID=your_agent_id
   

AI Interface Integration

To configure the server for Cursor IDE or Claude Desktop:

1. Find the uv path:

which uv  # Or locate uv

2. Create and edit the config.json file and add the following:

   {
     "server": {
       "host": "localhost",
       "port": 8000
     },
     "tools": {
       "tool1": {
         "command": "python tool.py"
       }
     }
   }
   

Environment Setup

Use uv for dependency management:

uv init --python
uv run myapp

💻 Usage Examples

Basic Usage

In Cursor, an example question:

# Example question
server.get_current_time()

Server Processing Flow

1. Receive the query request.
2. Parse and execute the tool function.
3. Return the result.

Development

Modifying the Server

To add new features:

1. Add a new tool by creating an additional function and using the @mcp.tool() decorator:

   @mcp.tool()
   def new_tool(param: str) -> str:
       """Describe the function of the tool"""
       # Implementation code
       return result
   

Limitations

• The server runs locally and may not be available in a remote development environment.
• Tool responses are affected by Contextual AI API limitations and quotas.
• Currently, only the stdio transmission mode is supported.

For all the features of Contextual AI, please refer to the official documentation.