qdrant/mcp-server-qdrant

Overview

An official Model Context Protocol (MCP) server for Qdrant vector database. It enables AI agents to perform semantic search, store and retrieve vector embeddings, and manage collections through the standardized MCP interface. Written in Python, it provides a direct bridge between MCP-compatible clients and Qdrant's vector storage capabilities.

Best for

Best for
Developers building MCP-compatible AI agents that need vector search and memory via Qdrant

Use cases

• Building AI agents that need long-term memory via vector embeddings
• Enabling semantic search over documents or knowledge bases from an MCP client
• Integrating Qdrant vector storage into MCP-based agent workflows

How to use

Install

npx @smithery/cli install mcp-server-qdrant --client claude

Tools exposed

• information
• metadata
• collection_name
• QDRANT_URL
• QDRANT_API_KEY
• COLLECTION_NAME
• QDRANT_LOCAL_PATH
• EMBEDDING_PROVIDER
• EMBEDDING_MODEL
• TOOL_STORE_DESCRIPTION
• TOOL_FIND_DESCRIPTION
• QDRANT_SEARCH_LIMIT
• QDRANT_READ_ONLY
• FASTMCP_LOG_LEVEL
• FASTMCP_SERVER_DEBUG
• FASTMCP_SERVER_HOST
• FASTMCP_SERVER_PORT
• FASTMCP_SERVER_ON_DUPLICATE_RESOURCES
• FASTMCP_SERVER_ON_DUPLICATE_TOOLS
• FASTMCP_SERVER_ON_DUPLICATE_PROMPTS

Tested with

Claude Desktop, Claude Code, Cursor, Windsurf, VS Code

Notes

An official Model Context Protocol (MCP) server for Qdrant vector database. It enables AI agents to perform semantic search, store and retrieve vector embeddings, and manage collections through the standardized MCP interface. Written in Python, it provides a direct bridge between MCP-compatible clients and Qdrant’s vector storage capabilities.

1,420 stars on GitHub. Last updated 2026-05-20. Licensed Apache-2.0.

Use cases

• Building AI agents that need long-term memory via vector embeddings
• Enabling semantic search over documents or knowledge bases from an MCP client
• Integrating Qdrant vector storage into MCP-based agent workflows

Pros

• Official implementation maintained by Qdrant
• Leverages the MCP standard for interoperability with many AI clients
• Python-based, easy to extend or embed in existing Python projects

Cons

• Requires a running Qdrant instance (local or cloud) to function
• Limited to the MCP protocol; not a general-purpose Qdrant client
• Setup and configuration may be non-trivial for beginners

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.