MCP Server with RAG and Multi-Search

What is MCP Server with RAG and Multi-Search?

This is a custom Model Context Protocol (MCP) server that provides Retrieval-Augmented Generation (RAG) capabilities using LlamaIndex and multiple web search options via Google's Gemini 2.0 API and Linkup. It is intended for developers who want to combine local document querying with AI-powered web search.

How to use MCP Server with RAG and Multi-Search?

Clone the repository, install dependencies with pip install -r requirements.txt, set up a .env file with your GEMINI_API_KEY (required) and optionally LINKUP_API_KEY, and add documents to the data directory. Then start the server with python server.py. The server exposes four tools: web_search, gemini_search, linkup_search, and rag.

Key features of MCP Server with RAG and Multi-Search

• RAG workflow using local documents with LlamaIndex
• Multiple web search options (Gemini 2.0 and Linkup)
• Built with FastMCP for easy integration
• Local LLM integration via Ollama (DeepSeek models)

Use cases of MCP Server with RAG and Multi-Search

• Query your own documents while also searching the web for supplementary information
• Build a retrieval-augmented chatbot that can answer from private data and public sources
• Compare or fallback between different search providers (Gemini AI vs Linkup)
• Experiment with local RAG pipelines using a lightweight MCP server

FAQ from MCP Server with RAG and Multi-Search

What dependencies are required to run this server?

Python 3.8 or higher, Ollama installed locally with the DeepSeek model (pull deepseek-r1:1.5b), a Gemini API key, and optionally a Linkup API key.

How do I set up the server?

Clone the repository, install requirements, create a .env file with your API keys, add documents to the data folder, then start the server with python server.py.

What models are used for local RAG?

The server uses Ollama with DeepSeek models by default. The README notes you can modify the code to use a different model.

Where are my local documents stored?

Place your documents in the data directory at the project root. This folder is created automatically if it doesn’t exist.

What should I do if I encounter Python 3.13 compatibility issues?

The README recommends downgrading to Python 3.11 or 3.10 if you face issues with Python 3.13. Also ensure Ollama is running and API keys are correct.