Qdrant MCP Server

🚀 Qdrant MCP Server

A MCP server for interacting with the Qdrant vector database. This server provides tools for managing vectors, performing similarity searches, and automatically using text-to-vector embeddings (through the MCP (Master Control Program) framework).

🚀 Quick Start

The Qdrant MCP Server is designed to facilitate seamless interaction with the Qdrant vector database. It offers a range of features for managing vectors and performing similarity searches.

✨ Features

• Automatically perform text-to-vector embeddings using FastEmbed.
• Store and retrieve text content with vector search capabilities.
• Configure the default collection via environment variables.
• Conduct text similarity searches based on content.
• Efficiently embed using optimized models.

📦 Installation

Configuration

Create a .env file based on the .env.example template:

# Qdrant connection settings
QDRANT_HOST=localhost
QDRANT_PORT=6333
QDRANT_API_KEY=
QDRANT_VERIFY_SSL=True  # Set to False if using a self-signed certificate

# Default settings
DEFAULT_COLLECTION_NAME=default_collection
EMBEDDING_MODEL=BAAI/bge-small-en-v1.5

You can change the embedding model to any model supported by FastEmbed.

Local Run

1. Install the package:

pip install -e .

2. Run the server:

qdrant-mcp-server

Docker Run

1. Build the Docker image:

docker build -t qdrant-mcp-server .

2. Run the container:

docker run -p 8000:8000 --env QDRANT_HOST=<your-qdrant-host> --env QDRANT_PORT=<your-qdrant-port> --env QDRANT_VERIFY_SSL=<True|False> qdrant-mcp-server

💻 Usage Examples

Basic Usage

Store Text

await store_text(
    text="What is the capital of France?", 
    metadata={"category": "geography", "type": "question"}
)

Search Similar Text

await search_similar_text(
    query="What is Paris the capital of?",
    limit=5
)

Batch Store Texts

await store_texts(
    texts=["Paris is in France", "London is in England", "Berlin is in Germany"],
    metadatas=[
        {"category": "geography", "country": "France"},
        {"category": "geography", "country": "England"},
        {"category": "geography", "country": "Germany"}
    ]
)

📚 Documentation

Testing

This package includes a test suite to verify functionality. To run the tests:

1. Install development dependencies:

pip install -e ".[dev]"

2. Run the tests:

cd tests
./run_tests.py

Or directly use pytest:

pytest -xvs tests/

Using a Self-Signed Certificate

If your Qdrant server uses a self-signed certificate, set QDRANT_VERIFY_SSL=False in the .env file or when running the Docker container. This disables SSL certificate verification.

Tools

The server provides the following tools:

Text Tools

• store_text: Convert text to an embedding vector and store it in the database.
• search_similar_text: Convert the query text to an embedding and find similar vectors.
• store_texts: Batch convert multiple texts to embeddings and store them.

Vector Tools

• search_vectors: Search for similar vectors in a collection.
• upsert_vectors: Upload vectors to a collection.
• filter_search: Perform a search using metadata filters.

Point Tools

• get_points: Retrieve points by ID from a collection.
• delete_points: Delete points by ID from a collection.
• count_points: Count the number of points in a collection.