MCP with RAG Demo
About
This demonstration project shows how to implement a Model Context Protocol (MCP) server with Retrieval-Augmented Generation (RAG) capabilities. The demo allows AI models to interact with a knowledge base, search for information, and add new documents.
Details
- License
- MIT license
Explore
- MCP server with tool and resource support
- RAG implementation (with fallback to in-memory storage)
- Client example for interacting with the MCP server
- Support for both SSE (HTTP) and stdio communication modes
- Simple prompt templates
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
MCP with RAG DemoCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
- Python 3.8+
- pip (Python package manager)
source mcp-env/bin/activate # On Windows: mcp-env\Scripts\activate
The following environment variables can be set in the .env file:
- OPENAI_API_KEY: Your OpenAI API key (required)
- MCP_SERVER_URL: The URL of the MCP server (optional)
- DEBUG: Enable debug mode (optional)
Run the example script:
python openai_example.py
You can also specify options directly:
python openai_example.py --api-key your_api_key_here --server-url http://localhost:3000 --debug
echo
A simple tool that echoes back the provided message
add
A tool that adds two numbers together
add_document
Adds a document to the knowledge base
rag_search
Searches the knowledge base for information related to a query
list_documents
Lists all documents in the knowledge base
The following tools are available in this demo:
- echo: A simple tool that echoes back the provided message
- add: A tool that adds two numbers together
- add_document: Adds a document to the knowledge base
- rag_search: Searches the knowledge base for information related to a query
- list_documents: Lists all documents in the knowledge base
To add new tools, modify the server.py file:
```python
@mcp.tool()
def your_tool_name(param1: type, param2: type) -> return_type:
"""Tool description"""
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"mcp with rag demo": {
"mcp-demo-thinklytics": {
"command": "python",
"args": [
"-m",
"venv",
"mcp-env"
]
}
}
}
}McpServers
{
"mcp-demo-thinklytics": {
"command": "python",
"args": [
"-m",
"venv",
"mcp-env"
]
}
}This demonstration project shows how to implement a Model Context Protocol (MCP) server with Retrieval-Augmented Generation (RAG) capabilities. The demo allows AI models to interact with a knowledge base, search for information, and add new documents.
Features
- MCP server with tool and resource support
- RAG implementation (with fallback to in-memory storage)
- Client example for interacting with the MCP server
- Support for both SSE (HTTP) and stdio communication modes
- Simple prompt templates
Prerequisites
- Python 3.8+
- pip (Python package manager)
Installation
1. Clone the repository:
git clone <repository-url>
cd mcp-demo
2. Create a virtual environment:
python -m venv mcp-env
source mcp-env/bin/activate # On Windows: mcp-env\Scripts\activate
3. Install the required dependencies:
pip install -r requirements.txt
Project Structure
mcp-demo/
├── server.py # Main MCP server implementation
├── client_example.py # Example client to interact with the server
├── requirements.txt # Project dependencies
├── sample_data.txt # Sample data available as a resource
├── tools/
│ ├── __init__.py # Package initialization
│ └── rag_tools.py # RAG tools implementation
└── README.md # This readme file
Running the Demo
Step 1: Start the MCP Server
You can run the MCP server in two different modes:
Option A: SSE (HTTP) Mode
This mode allows the server to accept connections over HTTP using Server-Sent Events (SSE):
python server.py --sse
By default, the server will listen on 0.0.0.0:8000. You can customize the host and port:
python server.py --sse --host 127.0.0.1 --port 9000
Option B: stdio Mode
This mode allows the server to communicate through standard input/output:
python server.py --stdio
Step 2: Run the Client Example
After the server is up and running, open a new terminal window (keeping the server running in the first one):
Connecting to an SSE Server
```bash
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
