百度向量数据库MCP Server

What is 百度向量数据库MCP Server?

A Model Context Protocol (MCP) server that provides access to Baidu Cloud Vector Database functionality. It allows large language model applications that support MCP to directly interact with a Baidu Cloud vector database for tasks like listing databases, managing tables, performing vector similarity searches, and full-text searches.

How to use 百度向量数据库MCP Server?

Clone the repository and run the server via uv with the required --endpoint and --api-key arguments. Alternatively, set the MOCHOW_ENDPOINT and MOCHOW_API_KEY environment variables in a .env file inside the source directory. The server can be used with Claude Desktop, Cursor, or any custom MCP client by configuring the MCP connection to point to the server's executable path.

Key features of 百度向量数据库MCP Server

• List, create, and switch between databases
• Describe table schemas and view table statistics
• Insert and delete rows using filter expressions
• Create, rebuild, and drop vector indexes
• Execute vector similarity search with scalar filtering
• Perform full‑text search with filtering and limiting

Use cases of 百度向量数据库MCP Server

• AI‑powered code editors querying vector embeddings stored in Baidu Cloud
• Chat‑based assistants retrieving relevant documents via vector search
• Automated database management tasks (e.g., listing tables, creating indexes)
• Building RAG (Retrieval‑Augmented Generation) pipelines with a cloud vector database

FAQ from 百度向量数据库MCP Server

What are the runtime requirements?

Python 3.10 or higher must be installed, along with the uv package runner.

How do I configure the server with Claude Desktop?

Edit the Claude Desktop configuration file (claude_desktop_config.json) to add the server under mcpServers, specifying the full path to uv and the server script with the endpoint and API key arguments.

What applications can use this MCP server?

Any application that supports the MCP protocol, such as Claude Desktop and Cursor, can connect to it.

Where does the data live?

All data resides in the user’s Baidu Cloud Vector Database instance. The server acts as a bridge, forwarding queries and results.

What transport does the server use?

The server uses standard MCP transport (stdio). It does not expose HTTP endpoints; communication is handled via stdin/stdout of the child process.