Qdrant - MemWire

Qdrant is the default vector store for MemWire. Three deployment options are supported: embedded (no server), local Docker, or managed Qdrant Cloud.

Embedded mode

The simplest option. Qdrant runs in-process and stores data in a local directory. No Docker or external service required.

from memwire import MemWire, MemWireConfig

config = MemWireConfig(
    qdrant_path="./memwire_data",       # directory to store vector data
    qdrant_collection_prefix="app_",   # optional prefix for collection names
)
memory = MemWire(config=config)

Use qdrant_path=":memory:" in tests for a fast in-memory store that resets between runs.

Local Docker server

Run Qdrant as a container and connect over HTTP.

docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant

config = MemWireConfig(
    qdrant_url="http://localhost:6333",
    qdrant_collection_prefix="app_",
)
memory = MemWire(config=config)

Or spin up Qdrant alongside the MemWire REST API using the provided docker-compose.yml:

cd api
docker compose up

Qdrant Cloud

Sign up at cloud.qdrant.io, create a cluster, and copy your cluster URL and API key.

config = MemWireConfig(
    qdrant_url="https://<cluster-id>.us-east4-0.gcp.cloud.qdrant.io",
    qdrant_api_key="your-qdrant-api-key",
    qdrant_collection_prefix="prod_",
)
memory = MemWire(config=config)

Configuration reference

Parameter	Default	Description
`qdrant_url`	`None`	Qdrant server URL. Omit to use embedded mode.
`qdrant_api_key`	`None`	API key for Qdrant Cloud. Not required for local servers.
`qdrant_path`	`None`	Local directory for embedded storage. Use `":memory:"` for tests.
`qdrant_collection_prefix`	`""`	Prefix applied to all collection names. Useful for multi-tenant isolation.