mcp-server-milvus

Category: Database Messaging MCP Servers
Source: GitHub Repository

MCP server for Milvus/Zilliz vector databases, enabling direct interaction with your database through the Model Context Protocol (MCP). Provides a standardized way to connect LLM (Large Language Model) applications with Milvus vector database features.

Features

General

• Implements the Model Context Protocol (MCP) to enable integration between LLM applications and Milvus vector databases.
• Can be used with various LLM-enabled applications such as Claude Desktop, Cursor, and any custom MCP client.
• Runs with Python 3.10+ and requires a running Milvus instance.
• Can be run directly via uv without separate installation.
• Supports configuration via environment variables (MILVUS_URI, MILVUS_TOKEN, MILVUS_DB).

Available Tools

Search and Query Operations

• milvus-text-search: Full-text document search in collections.
• milvus-vector-search: Vector similarity search using different distance metrics (COSINE, L2, IP).
• milvus-hybrid-search: Combines vector similarity and attribute-based filtering.
• milvus-multi-vector-search: Similarity search with multiple query vectors.
• milvus-query: Query collections using filter expressions.
• milvus-count: Count entities in a collection, with optional filtering.

Collection Management

• milvus-list-collections: Lists all collections in the database.
• milvus-collection-info: Retrieves detailed information about a collection.
• milvus-get-collection-stats: Provides statistics about a collection.
• milvus-create-collection: Creates a new collection with a specified schema and optional index parameters.
• milvus-load-collection: Loads collections into memory.
• milvus-release-collection: Releases collections from memory.
• milvus-get-query-segment-info: Retrieves information about query segments.
• milvus-get-collection-loading-progress: Shows loading progress of a collection.

Data Operations

• milvus-insert-data: Inserts data into a collection.
• milvus-bulk-insert: Batch data insertion for better performance.
• milvus-upsert-data: Upserts (insert or update) data in a collection.
• milvus-delete-entities: Deletes entities from a collection using a filter expression.
• milvus-create-dynamic-field: Adds a dynamic field to an existing collection.

Index Management

• milvus-create-index: Creates indexes on vector fields.
• milvus-get-index-info: Retrieves index information for a collection.

Integration Examples

• Claude Desktop: Integrate by updating the configuration file and running the server with the appropriate Milvus URI.
• Cursor: Add the MCP server via the settings UI or project-specific configuration for Composer's Agent feature.

Development & Deployment

• Can be run locally or remotely.
• Provides troubleshooting steps for connection and authentication issues.

Pricing

No pricing information is provided; the project is open source and available on GitHub.

Tags

mcp, milvus, vector-database, database, ai-integration