qdrant/mcp-server-qdrant

What it solves

This project provides a standardized way for Large Language Models (LLMs) to interact with Qdrant, a vector search engine. It allows AI applications to use Qdrant as a semantic memory layer, enabling them to store and retrieve information based on meaning rather than just keywords.

How it works

It implements the Model Context Protocol (MCP), acting as a server that exposes specific tools to an MCP-compatible client (like Claude Desktop, Cursor, or VS Code). The server provides two primary functions:

• qdrant-store: Saves text information and optional metadata into a specified Qdrant collection.
• qdrant-find: Performs a semantic search to retrieve the most relevant information from the database based on a natural language query.

It uses the fastembed library for creating embeddings, with sentence-transformers/all-MiniLM-L6-v2 as the default model.

Who it’s for

• AI Developers: Those building LLM-powered applications that need a persistent, searchable memory.
• Software Engineers: Developers using AI-integrated IDEs (like Cursor or Windsurf) who want to create a semantic search tool for their own code snippets and documentation.

Highlights

• Multi-Transport Support: Supports stdio, sse (Server-Sent Events), and streamable-http for both local and remote connections.
• Flexible Deployment: Can be run via uvx, Docker, or integrated directly into Claude Desktop and VS Code.
• Customizable: Tool descriptions can be modified via environment variables to change how the AI perceives and uses the storage and retrieval tools.