qdrant

A plugin that provides vector similarity search capabilities using Qdrant vector database.

What it does

This plugin provides three main functionalities:

1. Create collections with configurable vector dimensions
2. Store documents with their vector embeddings in Qdrant
3. Search for similar documents using vector embeddings

Configuration

The plugin requires the following configuration:

{
  "plugins": [
    {
      "name": "qdrant",
      "path": "oci://ghcr.io/hyper-mcp-rs/qdrant-plugin:latest",
      "runtime_config": {
        "QDRANT_URL": "http://localhost:6334",
        "allowed_hosts": [
          "localhost:6333"
        ],
        "env_vars": {
          "QDRANT_URL": "http://localhost:6333"
        }
      }
    }
  ]
}

Tools

1. qdrant_create_collection

Creates a new collection in Qdrant with specified vector dimensions.

{
  "collection_name": "my_documents",
  "vector_size": 384  // Optional, defaults to 384
}

2. qdrant_store

Stores a document with its vector embedding in Qdrant.

{
  "collection_name": "my_documents",
  "text": "Your document text",
  "vector": [0.1, 0.2, ...] // Vector dimensions must match collection's vector_size
}

3. qdrant_find

Finds similar documents using vector similarity search.

{
  "collection_name": "my_documents",
  "vector": [0.1, 0.2, ...],    // Vector dimensions must match collection's vector_size
  "limit": 5                    // Optional, defaults to 5
}

Features

• Configurable vector dimensions per collection
• Support for vector-based queries
• Configurable similarity search results limit
• Uses cosine similarity for vector matching
• Thread-safe operations

Dependencies

• Qdrant for vector storage and similarity search
• UUID for document identification