MCP Server for Qdrant
A Machine Control Protocol (MCP) server for storing and retrieving information from a Qdrant vector database.
Features
- Store text information with optional metadata in Qdrant
- Semantic search for stored information
- FastEmbed integration for text embeddings
- Environment-based configuration
- Docker support
Installation
Using pip
pip install mcp-server-qdrant
From source
git clone https://github.com/your-org/mcp-server-qdrant.git
cd mcp-server-qdrant
make setup
Configuration
Configuration is done through environment variables. You can create a .env file based on the .env.example file:
cp .env.example .env
Edit the .env file to configure the server:
# Qdrant configuration
QDRANT_URL=http://localhost:6333
QDRANT_API_KEY=your-api-key
# Collection name
COLLECTION_NAME=memories
# Embedding provider configuration
EMBEDDING_PROVIDER=fastembed
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
Usage
Running locally
python -m mcp_server_qdrant.main
Or using the make command:
make run
Docker
docker-compose up
Tools
The MCP server provides the following tools:
qdrant-store
Stores information in the Qdrant database.
information: The text to store
metadata: Optional JSON metadata to associate with the text
qdrant-find
Searches for information in the Qdrant database using semantic search.
query: The search query
Development
Testing
make test
Formatting
make format
Linting
make lint
Building
make build
License
Apache License 2.0