Milvus MCP Server

Example Use Cases

• 1

  Perform vector similarity and hybrid searches against a Milvus collection directly from conversational workflows.

• 2

  Manage Milvus collections and indexes (create, load, release, get stats) programmatically via MCP.

• 3

  Ingest, upsert, batch-insert, or delete data in Milvus and run attribute-filtered queries for retrieval-augmented generation.

Description

This MCP server provides a set of tools that allow Model Context Protocol (MCP) clients (like Claude Desktop and Cursor) to interact with a Milvus vector database. It supports text, vector, hybrid and multi-vector searches, collection and index management, and data insertion/upsert/delete operations. The server runs as a standalone process (recommended with uv) and can be configured via command-line flags or environment variables (MILVUS_URI, MILVUS_TOKEN, MILVUS_DB). It enables seamless integration of Milvus functionality into conversational agents and custom MCP clients.