Qdrant

The Qdrant MCP server enables AI agents to perform semantic search and manage vector collections in Qdrant. Ideal for RAG applications, recommendation systems, and similarity search.

Capabilities

• Collections — Create, list, and manage vector collections
• Search — Perform semantic vector search with filters
• Upsert — Insert and update vector points with payloads
• Delete — Remove vectors by ID or filter conditions
• Payload — Read and update metadata payloads
• Health — Monitor collection health and statistics
• Snapshots — Create and restore collection snapshots

Usage

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

Configuration

1. Install Qdrant: docker run -p 6333:6333 qdrant/qdrant
2. Set QDRANT_URL environment variable (default: http://localhost:6333)
3. Optional: Set QDRANT_API_KEY for cloud instances