MCP Server Lancedb

🚀 mcp-lance-db: MCP Server for LanceDB

The Model Context Protocol (MCP) is an open protocol that enables seamless integration between LLM applications and external data sources and tools. Whether you're building an AI-powered IDE, enhancing a chat interface, or creating custom AI workflows, MCP provides a standardized way to connect LLMs with their required context.

This repository demonstrates how to create an MCP server for LanceDB, an embedded vector database.

📋 Overview

This is a basic Model Context Protocol server for storing and retrieving memories in the LanceDB vector database. It acts as a semantic memory layer, allowing the storage of text and its embedding vectors for later retrieval.

🧩 Components

Tools

The server implements two tools:

• add-memory: Adds a new memory to the vector database.

  • Accepts a required string parameter "content".
  • Stores the text for later retrieval via its embedding vector.

• search-memories: Retrieves semantically similar memories.

  • Accepts a required string parameter "query".
  • An optional "limit" parameter to control the number of results (default value is 5).
  • Returns memories ranked by semantic similarity to the query.
  • Updates the server state and notifies the client of resource changes.

⚙️ Configuration

The server uses the following configuration:

• Database path: "./lancedb"
• Collection name: "memories"
• Embedding provider: "sentence-transformers"
• Model: "BAAI/bge-small-en-v1.5"
• Device: "cpu"
• Similarity threshold: 0.7 (upper bound of the distance range)

🚀 Quick Start

Running on Claude Desktop

On MacOS: ~/Library/Application\ 支持/Claude/claude_desktop_config.json On Windows: %APPDATA%/Claude/claude_desktop_config.json%

Configure as follows:

{
  "lancedb": {
    "command": "uvx",
    "args": [
      "mcp-lance-db"
    ]
  }
}

💻 Development

Build and Publish

To prepare the distribution package:

1. Synchronize dependencies and update the lock file:

uv sync

2. Build the package release:

uv build

This will create source code and wheel format releases in the dist/ directory.

3. Publish to PyPI:

uv publish

Note: You need to set PyPI credentials via environment variables or command-line parameters:

• Token: --token or UV_PUBLISH_TOKEN
• Or username/password: --username/UV_PUBLISH_USERNAME and --password/UV_PUBLISH_PASSWORD

Debugging

Since the MCP server runs via standard input and output, debugging can be challenging. For the best debugging experience, we highly recommend using MCP Inspector.

You can start the MCP Inspector with the following command:

npx @modelcontextprotocol/inspector uv --directory $(PWD) run mcp-lance-db

After starting, the Inspector will display a URL that you can open in your browser to start debugging.