DuckDB RAG MCP Sample

This is a sample that embeds and vectorizes a markdown document so that it can be explained using MCP and RAG.

We use Plamo-Embedding-1B for vectorization.

function

Extract and vectorize text from markdown files
Vector Searching with DuckDB
Persisting vector data with Parquet files
Vector search from MCP

How to use

Vector data generation

First, place the markdown files you want to search in a specific directory, then convert them to Parquet files with the following command.

uv run main.py --directory ~/path/to/markdown/files --parquet vectors.parquet

Configuring MCP

Build

The following command will generate a single binary in dist/server .

uv run pyinstaller --clean --strip --noconfirm --onefile server.py

MCP Client Configuration

Please set it according to the client you want to use.

For Claude Desktop it looks like this:

For VECTOR_PARQUET, specify the file you just converted.

uv run mcp install server.py -v VECTOR_PARQUET=/path/to/vectors.parquet

It is set as follows:

{
  "mcpServers": {
    "DuckDB-RAG-MCP-Sample": {
      "command": "/path/to/dist/server",
      "env": {
        "VECTOR_PARQUET": "/path/to/vectors.parquet"
      }
    }
  }
}

Start the development server

uv run mcp dev server.py

license

The DuckDB RAG MCP Sample is provided under the Apache License, Version 2.0.

Integrations