mcp-server-qdrant

608

The mcp-server-qdrant project is designed to integrate Qdrant, a vector search engine, with the Model Context Protocol to provide a robust semantic memory layer. It allows users to store and retrieve data across a Qdrant database using standardized MCP tools, making it suitable for various AI applications.

Overview

  • The project is an official Model Context Protocol (MCP) server for Qdrant, functioning as a semantic memory layer over the Qdrant vector search engine.
  • It provides tools for storing (qdrant-store) and retrieving (qdrant-find) information within the Qdrant database.

Components

  • qdrant-store: Stores information with optional metadata in specified collections.
  • qdrant-find: Retrieves relevant information from specified collections based on queries.

Installation

  • Supports installation and operation using uvx, Docker, and Smithery.

Usage

  • The server can be configured using environment variables or command-line tools.
  • Supports various transport protocols, including stdio and sse, with options for local or remote connections.

Integration

  • Compatible with MCP-compatible clients like Cursor and VS Code, providing support for semantic code search and storage.