mcp-agentic-rag

What is mcp-agentic-rag?

mcp-agentic-rag is an MCP server and client implementation for building agentic Retrieval-Augmented Generation (RAG) applications. It exposes tools that enhance RAG performance, such as entity extraction, query refinement, and relevance checking.

How to use mcp-agentic-rag?

Clone the repository, install dependencies (pip install -r requirements.txt), configure the OPENAI_MODEL_NAME environment variable in a .env file, then start the server with python server.py and run the client with python mcp-client.py.

Key features of mcp-agentic-rag

• Exposes tools for entity extraction, query refinement, and relevance checking.
• Uses OpenAI models for LLM-based operations.
• Built with FastMCP from the mcp library.
• Client demonstrates connection, tool listing, and tool invocation.
• Supports environment-based configuration via .env file.

Use cases of mcp-agentic-rag

• Improve document retrieval by extracting key entities from user queries.
• Enhance query quality before retrieval with LLM-based refinement.
• Filter irrelevant documents by checking relevance of text chunks to the question.
• Build a complete agentic RAG pipeline with tool orchestration.

FAQ from mcp-agentic-rag

What are the dependencies of mcp-agentic-rag?

The server requires Python 3.7+ and the packages openai, mcp, and dotenv.

How do I configure the server?

Create a .env file based on the provided .env.sample and set the OPENAI_MODEL_NAME environment variable to the desired OpenAI model.

What tools does the server provide?

The server provides four tools: get_time_with_prefix, extract_entities_tool, refine_query_tool, and check_relevance.

How do I start and interact with the server?

Start the server with python server.py. Run the client with python mcp-client.py to connect, list tools, and call them with arguments.

Does the server require an OpenAI API key?

Yes, the tools that use OpenAI (entity extraction, query refinement, relevance checking) require the OPENAI_MODEL_NAME environment variable to be set, which implies an OpenAI API key must be configured in the environment or .env file.