Qdrant Retrieve MCP Server

@gergelyszerovay

リポジトリ

基本情報

カテゴリ

データベース

ライセンス

MIT license

トランスポート

stdio

作者

gergelyszerovay

Qdrant Retrieve MCP Server について

MCP server for semantic search with Qdrant vector database

#qdrant #semantic-search #mcp-server

Qdrant Retrieve MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "qdrant": {
      "command": "npx",
      "args": [
        "-y",
        "@gergelyszerovay/mcp-server-qdrant-retrive"
      ],
      "env": {
        "QDRANT_API_KEY": "your_api_key_here"
      }
    }
  }
}

Qdrant Retrieve MCP Server のツール

Retrieves semantically similar documents from multiple Qdrant vector store collections based on multiple queries

概要

What is Qdrant Retrieve MCP Server?

An MCP server that enables semantic search across multiple Qdrant vector database collections. It supports multi-query input, configurable result counts, and collection source tracking, connecting to any Qdrant instance specified by URL.

How to use Qdrant Retrieve MCP Server?

Configure it in Claude Desktop by adding an MCP server entry with the command npx -y @gergelyszerovay/mcp-server-qdrant-retrive and optionally set the QDRANT_API_KEY environment variable. Alternatively, run it from the command line using flags like --qdrantUrl, --enableHttpTransport, or --embeddingModelType.

Key features of Qdrant Retrieve MCP Server

Semantic search across multiple Qdrant collections
Multi-query support (array of query texts)
Configurable result count (topK)
Collection source tracking in results
Stdio and HTTP transport options
Optional REST API server

Use cases of Qdrant Retrieve MCP Server

Retrieve semantically similar documents from multiple Qdrant collections in one request
Integrate semantic search into AI assistants via the MCP protocol
Power applications that need multi-query document retrieval with similarity scores
Combine vector search with LLM tools for knowledge retrieval

よくある質問

What embedding model does the server use?

The default embedding model is `Xenova/all-MiniLM-L6-v2`. It can be customized via the `--embeddingModelType` command-line option.

How do I connect to my Qdrant instance?

Set the Qdrant URL using the `--qdrantUrl` option (default: `http://localhost:6333`). If authentication is required, provide an API key via the `QDRANT_API_KEY` environment variable.

Why is the first retrieval slow?

The first retrieve may be slower because the server downloads the required embedding model on initial use.

Can the server run over HTTP instead of stdio?

Yes, enable HTTP transport with the `--enableHttpTransport` flag. The MCP HTTP server port defaults to 3001.

What transport methods are available?

The server supports stdio transport (enabled by default), HTTP transport (optional), and a REST API server (optional, default port 3002).

「データベース」の他のコンテンツ

Elasticsearch MCP Server

elastic

685

MCP Alchemy

runekaagaard

A MCP (model context protocol) server that gives the LLM access to and knowledge about relational databases like SQLite, Postgresql, MySQL & MariaDB, Oracle, and MS-SQL.

412