Qdrant

v1.0.0Databasesstable

MCP server for document ingestion and semantic search on Qdrant. Enables ingesting local documents, generating embeddings with OpenAI, and performing vector search with metadata filters.

claudecursorllmmcpmcp-server
Share:
1,404
Stars
0
Downloads
0
Weekly
0/5

What is Qdrant?

Qdrant is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp server for document ingestion and semantic search on qdrant. enables ingesting local documents, generating embeddings with openai, and performing vector search with metadata filters.

MCP server for document ingestion and semantic search on Qdrant. Enables ingesting local documents, generating embeddings with OpenAI, and performing vector search with metadata filters.

This server falls under the Databases category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • MCP server for document ingestion and semantic search on Qdr

Use Cases

Vector search and semantic embedding
Local document ingestion and retrieval
qdrant

Maintainer

LicenseApache 2.0
Languagepython
Versionv1.0.0
UpdatedMay 20, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y mcp-server-qdrant

Manual Installation

npx -y mcp-server-qdrant

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Qdrant

The mcp-server-qdrant MCP server connects your AI assistant to a Qdrant vector database, enabling persistent semantic memory storage and retrieval across sessions. It exposes two core tools—qdrant-store and qdrant-find—that let the AI embed and index information using local FastEmbed or OpenAI-compatible models, then search it with natural language queries. Developers use this server to build long-term memory for agents, create searchable code-snippet libraries, and enable semantic retrieval over documentation without maintaining a separate search pipeline.

Prerequisites

  • Python 3.10+ with uvx (via uv) installed, or Docker for container-based deployment
  • A running Qdrant instance: either a local server at http://localhost:6333, or a Qdrant Cloud cluster URL and API key
  • An MCP client such as Claude Desktop, Cursor, or Windsurf
  • Optional: an OpenAI-compatible embedding model API key if you prefer cloud embeddings over the default local FastEmbed model
1

Install or start Qdrant

Run a local Qdrant instance via Docker or sign up for Qdrant Cloud. The MCP server needs a reachable Qdrant endpoint before it can store or search vectors.

docker run -p 6333:6333 qdrant/qdrant
2

Install mcp-server-qdrant via uvx

The server is distributed as a Python package on PyPI and is best run with uvx, which creates an isolated environment automatically. No global install is required.

uvx mcp-server-qdrant
3

Set required environment variables

Configure the server by setting environment variables. QDRANT_URL points to your Qdrant instance. COLLECTION_NAME is the vector collection to use (it will be created automatically if it does not exist). EMBEDDING_MODEL selects the local FastEmbed model.

export QDRANT_URL="http://localhost:6333"
export COLLECTION_NAME="my-collection"
export EMBEDDING_MODEL="sentence-transformers/all-MiniLM-L6-v2"
4

Add the server to your MCP client configuration

Edit your Claude Desktop config file (claude_desktop_config.json) to register the server. Pass environment variables directly in the config so the server starts with the correct settings each time.

5

Verify the connection

Open your MCP client and ask it to store a test note. The qdrant-store tool should succeed without errors, confirming the server can reach your Qdrant instance and create the collection.

Qdrant Examples

Client configuration (local Qdrant)

Claude Desktop configuration using uvx with a local Qdrant instance and the default FastEmbed model.

{
  "mcpServers": {
    "qdrant": {
      "command": "uvx",
      "args": ["mcp-server-qdrant"],
      "env": {
        "QDRANT_URL": "http://localhost:6333",
        "COLLECTION_NAME": "my-collection",
        "EMBEDDING_MODEL": "sentence-transformers/all-MiniLM-L6-v2"
      }
    }
  }
}

Client configuration (Qdrant Cloud)

Claude Desktop configuration pointing to a Qdrant Cloud cluster with API key authentication.

{
  "mcpServers": {
    "qdrant": {
      "command": "uvx",
      "args": ["mcp-server-qdrant"],
      "env": {
        "QDRANT_URL": "https://xyz-example.eu-central.aws.cloud.qdrant.io:6333",
        "QDRANT_API_KEY": "your_api_key_here",
        "COLLECTION_NAME": "project-memory",
        "EMBEDDING_MODEL": "sentence-transformers/all-MiniLM-L6-v2"
      }
    }
  }
}

Prompts to try

Example prompts that use the qdrant-store and qdrant-find tools to build semantic memory.

- "Remember this architecture decision: we use event sourcing for the orders service because it simplifies auditing."
- "Search my notes for anything about the orders service architecture."
- "Store this code snippet for parsing ISO dates in Python: [paste snippet]. Tag it as 'python, datetime'."
- "Find all code snippets related to date parsing."
- "Remember that the staging API endpoint is https://api-staging.example.com and the prod endpoint is https://api.example.com."
- "What API endpoints have I stored?"

Troubleshooting Qdrant

Connection refused error when the server tries to reach Qdrant

Confirm your Qdrant container or cloud instance is running and that QDRANT_URL is correct. For Docker, ensure the port mapping -p 6333:6333 is in place and no firewall blocks the port.

Collection not found or embedding dimension mismatch after changing EMBEDDING_MODEL

Changing the embedding model changes the vector dimension. Delete the existing collection in Qdrant (or use a new COLLECTION_NAME) so a fresh collection with the correct dimension is created on the next store operation.

uvx command not found

Install uv first with 'curl -LsSf https://astral.sh/uv/install.sh | sh', then reload your shell. uvx is bundled with uv and provides isolated Python tool execution.

Frequently Asked Questions about Qdrant

What is Qdrant?

Qdrant is a Model Context Protocol (MCP) server that mcp server for document ingestion and semantic search on qdrant. enables ingesting local documents, generating embeddings with openai, and performing vector search with metadata filters. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Qdrant?

Install via npm with the command: npx -y mcp-server-qdrant. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Qdrant?

Qdrant works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Qdrant free to use?

Yes, Qdrant is open source and available under the Apache 2.0 license. You can use it freely in both personal and commercial projects.

Browse More Databases MCP Servers

Explore all databases servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "qdrant": { "command": "npx", "args": ["-y", "mcp-server-qdrant"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Qdrant?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides