mcp-local-rag logo

mcp-local-rag

by nkapila6

mcp-local-rag is a "primitive" RAG-like web search model context protocol (MCP) server that runs locally. It provides web search capabilities to models without relying on external APIs.

View on GitHub

Last updated: N/A

What is mcp-local-rag?

mcp-local-rag is a local server that enables Retrieval-Augmented Generation (RAG) functionality for language models. It allows models to access and incorporate information from web searches into their responses without using external APIs.

How to use mcp-local-rag?

  1. Install uv. 2. Clone the GitHub repository (optional, can be skipped by directly pasting the configuration into Claude config). 3. Add the provided configuration to your Claude config, adjusting the directory path if necessary. When prompted to fetch/lookup/search the web, the model will use the MCP server.

Key features of mcp-local-rag

  • Local execution

  • No API dependency

  • RAG-like web search

  • Integration with Claude

  • Uses uv

Use cases of mcp-local-rag

  • Answering questions with up-to-date information

  • Providing context from web searches to language models

  • Enabling RAG functionality in environments without external API access

  • Improving the accuracy and relevance of model responses

FAQ from mcp-local-rag

What is uv?

uv is a fast, modern Python package installer and resolver.

Do I need to clone the repository?

No, you can skip cloning by directly pasting the configuration into your Claude config.

Where can I find the Claude configuration paths?

You can find the configuration paths here: https://modelcontextprotocol.io/quickstart/user

What Python version should I use?

The example configuration uses Python 3.10.

How does this improve model responses?

By providing real-time information from web searches, the model can answer questions more accurately and with more relevant context.