Mcp Local Rag

What is Mcp Local Rag about?

Mcp Local Rag enables an LLM to perform live web searches, compute semantic similarity between the query and fetched results, and retrieve concise markdown excerpts from the top‑ranked pages. It runs entirely locally and communicates via the Model Context Protocol (MCP).

How to use Mcp Local Rag?

1. Add an MCP server entry to your client configuration, pointing to the mcp-local-rag server.
2. Choose a launch method:
   • uvx (recommended for quick testing): install uv and run the command shown in the configuration snippet.
   • Docker: pull the image ghcr.io/nkapila6/mcp-local-rag:latest and execute the provided docker run command.
3. When the LLM needs up‑to‑date information, it calls the mcp-local-rag tool. The server performs the search, processes the results, and returns the markdown context to the model, which then generates the final response.

Key Features of Mcp Local Rag

• Live DuckDuckGo search retrieving the top 10 results.
• Embedding generation using Google's MediaPipe Text Embedder.
• Cosine similarity ranking to select the most relevant entries.
• Automatic extraction of readable markdown from HTML pages.
• Completely local execution – no external API keys required.
• Simple MCP server integration; works with any MCP‑compatible client supporting tool calling.

Use Cases of Mcp Local Rag

• Supplying recent news or product releases to LLMs that lack up‑to‑date knowledge.
• Building RAG pipelines for internal knowledge bases without relying on cloud services.
• Enhancing chatbot experiences in environments with strict data residency requirements.
• Rapid prototyping of LLM tool‑calling workflows during development.

FAQ

Q: Do I need an API key? A: No. All operations run locally using public DuckDuckGo search and open‑source embedding models.

Q: Which LLM clients are compatible? A: Any client that implements the MCP tool‑calling interface, e.g., Claude Desktop, Cursor, Goose, and others.

Q: Can I adjust the number of search results or the embedding model? A: Yes. The source code is configurable; modify the search_limit or replace the MediaPipe embedder with another model as needed.

Q: How is the returned context formatted? A: The server extracts the main textual content from each URL and converts it to markdown before sending it back to the LLM.

Q: Is Docker required? A: Docker is optional but recommended for reproducible environments. The uvx method works directly on a machine with Python 3.10+ and uv installed.

mcp-local-rag

"primitive" RAG-like web search model context protocol (MCP) server that runs locally. ✨ no APIs ✨

Installation

Locate your MCP config path here or check your MCP client settings.

Run Directly via uvx

This is the easiest and quickest method. You need to install uv for this to work. Add this to your MCP server configuration:

{
  "mcpServers": {
    "mcp-local-rag":{
      "command": "uvx",
        "args": [
          "--python=3.10",
          "--from",
          "git+https://github.com/nkapila6/mcp-local-rag",
          "mcp-local-rag"
        ]
      }
  }
}

Using Docker (recommended)

Ensure you have Docker installed. Add this to your MCP server configuration:

{
  "mcpServers": {
    "mcp-local-rag": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "--init",
        "-e",
        "DOCKER_CONTAINER=true",
        "ghcr.io/nkapila6/mcp-local-rag:latest"
      ]
    }
  }
}

Security audits

MseeP does security audits on every MCP server, you can see the security audit of this MCP server by clicking here.

MCP Clients

The MCP server should work with any MCP client that supports tool calling. Has been tested on the below clients.

• Claude Desktop
• Cursor
• Goose
• Others? You try!

Examples on Claude Desktop

When an LLM (like Claude) is asked a question requiring recent web information, it will trigger mcp-local-rag.

When asked to fetch/lookup/search the web, the model prompts you to use MCP server for the chat.

In the example, have asked it about Google's latest Gemma models released yesterday. This is new info that Claude is not aware about.

Result

mcp-local-rag performs a live web search, extracts context, and sends it back to the model—giving it fresh knowledge:

Contributing

Have ideas or want to improve this project? Issues and pull requests are welcome!

License

This project is licensed under the MIT License.