MCPOfficialQdrant
Qdrant MCP
Stores information in Qdrant and finds it again with semantic search, as agent memory.
The official Qdrant server works as semantic memory: the qdrant-store tool saves text with metadata, and qdrant-find searches for similar entries. FastEmbed computes embeddings locally, and the connection to a server or a local database file is set in the QDRANT_URL or QDRANT_LOCAL_PATH variable; QDRANT_READ_ONLY turns off writes.
Installation
Manual install
$claude mcp add qdrant -e QDRANT_URL="http://localhost:6333" -e COLLECTION_NAME="<COLLECTION>" -- uvx mcp-server-qdrantDetails
- Works in
- Claude Code, Claude, Cursor, VS Code, Windsurf, Codex, Gemini CLI, Zed
- Transport
- local
- License
- Apache-2.0
- Tasks
- Data, Research
- Rating
- Source
- github.com/qdrant/mcp-server-qdrant (opens in a new tab)
Tags
- qdrant
- vector database
- memory
- semantic search
- embeddings
A pack is your own set of skills and MCP servers. sciver loads it in your AI with one command.