MCP Agentic Rag

🚀 MCP-agen-tic-RAG

This project implements an MCP (Model Context Protocol) server and client for building enhanced RAG (Retrieval Augmented Generation) applications. The server offers a set of tools to improve the performance of RAG systems, such as entity extraction, query optimization, and relevance checking. The client demonstrates how to connect to the server and use its tools.

🚀 Quick Start

Server Startup

python server.py

Client Execution

python mcp-client.py

✨ Features

Server Features

The server, based on the FastMCP class in the mcp library, provides the following tools:

• get_time_with_prefix: Returns the current date and time.
• extract_entities_tool: Uses OpenAI to extract entities from a given text query. This tool can be used to identify key entities in user queries, thereby improving the relevance of document retrieval.
• refine_query_tool: Uses OpenAI to optimize a given text query. This tool can be used to enhance the quality of user queries, thus improving the relevance of document retrieval.
• check_relevance: Uses an LLM to check the relevance of text segments to a given question. This tool can be used to filter out irrelevant content from retrieval results.

Client Features

The client demonstrates how to connect to the MCP server and use its tools. It establishes a connection to the server through the ClientSession class from the mcp library. The client provides the following examples:

• How to connect to the server
• Listing available tools
• Invoking specific tools and their parameters
• Processing queries using OpenAI and available MCP tools

📦 Installation

1. Clone the repository:

git clone https://github.com/rukshanet/mcp-agentic-rag.git

2. Install dependencies:

pip install -r requirements.txt

3. Configure environment variables:
   • Create a .env file based on the .env.sample file.
   • Set the OPENAI_MODEL_NAME environment variable to the name of the OpenAI model you want to use.

💻 Usage Examples

Basic Usage

The following example shows how to start the server and client:

# Start the server
python server.py
# Run the client
python mcp-client.py

📄 License

MIT