A Model Context Protocol (MCP) server for enhanced Qdrant vector database functionality. This server provides a tool for searching the Qdrant vector database.
-
qdrantSearch (default): Searches a Qdrant vector database collection — name and description can be overridden via
--name
and--description
flags
You can use it directly with npx
without global installation:
npx qdrant-search-mcp-server
When using npx
, the server will load configuration from a .env
file in the current directory if present. You can configure these environment variables:
QDRANT_URL=https://qdrant.is.solo.io
QDRANT_API_KEY=add your key
# As long as the embedding service is openai api compatible, you can use them also.
OPENAI_API_KEY=add your key
OPENAI_ENDPOINT=https://api.openai.com/v1
OPENAI_MODEL=text-embedding-3-large
Meanwhile, collections are provided directly in the command line, or the defaults will be used if nothing is specified.
npx qdrant-search-mcp-server --collections=istio,gloo-mesh-enterprise
Additionally, you can customize the MCP tool metadata at launch:
npx qdrant-search-mcp-server \
--collections=istio,gloo-mesh-enterprise \
--name=mySearchTool \
--description="Custom Qdrant search tool" # override default name/description
To use this MCP server with Claude, add it to your MCP settings configuration file:
{
"mcpServers": {
"qdrantSearch": {
"command": "npx",
"args": ["qdrant-search-mcp-server --collections=istio,gloo-mesh-enterprise --name=mySearchTool --description='Search Documentation related to Istio and Gloo Mesh'"],
"env": {
"QDRANT_URL": "http://localhost:6333",
"QDRANT_API_KEY": "your_api_key",
"OPENAPI_API_KEY": "your_api_key",
}
}
}
}
To use this MCP server with Visual Studio Code, you can add it to your mcp.json
configuration file. Below is an example configuration:
{
"servers": {
"qdrantSearch": {
"command": "npx",
"args": ["qdrant-search-mcp-server"],
"env": {
"QDRANT_URL": "http://localhost:6333",
"QDRANT_API_KEY": "your_api_key",
"OPENAPI_API_KEY": "your_api_key",
}
}
}
}
use_mcp_tool
server_name: qdrantSearch
tool_name: qdrantSearch
arguments: {
"query": "your search query",
"collection": "my-collection",
"limit": 5
}