MCP Milvus

Provides MCP-based Milvus vector database operations including database, collection, index management and vector search.

go

1

GitHub Stars

go

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "tailabs-mcp-milvus": {
      "command": "./build/mcp-milvus",
      "args": [],
      "env": {
        "PORT": "8080",
        "LOG_LEVEL": "info"
      }
    }
  }
}

You run an MCP Milvus server to manage Milvus databases, collections, and indexes through a standardized MCP interface. This server enables high-performance vector search workflows, session management, and middleware-ready deployment, including containerized runs, while keeping type-safe interactions.

How to use

You connect to the MCP Milvus server from an MCP client and perform operations using the available tools exposed by the server. Start by launching the server, then establish a connection to Milvus, and finally execute operations such as creating databases, managing collections and indexes, and performing vector searches. Use the client to issue requests that align with the lifecycle of Milvus resources and the data you store.

How to install

Prerequisites you need before installation include a working Go development environment and a running Milvus instance.

Install from source by cloning the project, installing dependencies, and building the binary.

# Using Git to clone the repository and build
git clone https://github.com/tailabs/mcp-milvus.git
cd mcp-milvus
make deps
make build

# Alternative build with Go directly
git clone https://github.com/tailabs/mcp-milvus.git
cd mcp-milvus
go mod download
go build -o mcp-milvus ./cmd/mcp-milvus

Run and connect

Start the server using the recommended runtime methods. You can either use the Makefile target to run the server, or run the built binary directly.

# Start via Makefile
make run

# Start the built binary directly
./build/mcp-milvus

Connect to Milvus and perform operations

After the server is running, you connect to Milvus using the built-in connector utility to establish the Milvus session. Provide the Milvus address, authentication token (if required), and the target database.

Available tools

milvus_create_database

Create a new Milvus database within the selected Milvus server instance.

milvus_list_databases

List all databases available in the connected Milvus server.

milvus_use_database

Switch the current context to a different Milvus database.

milvus_create_collection

Create a new collection within the selected Milvus database.

milvus_drop_collection

Delete an existing collection and its data from Milvus.

milvus_list_collections

List all collections in the current database.

milvus_get_collection_info

Retrieve metadata and schema details for a specific collection.

milvus_rename_collection

Rename an existing collection.

milvus_load_collection

Load a collection into memory for faster access.

milvus_release_collection

Release a loaded collection from memory.

milvus_create_index

Create an index on a collection to optimize queries.

milvus_drop_index

Remove an existing index from a collection.

milvus_insert_data

Insert new entities into a Milvus collection.

milvus_upsert

Insert or update entities in a Milvus collection.

milvus_delete_entities

Delete entities matching given criteria from a collection.

milvus_query

Execute conditional queries against a collection.

milvus_vector_search

Perform vector similarity searches against a collection.

milvus_connector

Establish a connection to a Milvus instance.