CLI reference
Console scripts: oneopen and oneopen-ml-studio (same entry point).
python -m oneopen_ml_studio <command>
Commands
oneopen start (recommended)
Start the local web server (Uvicorn).
oneopen start
oneopen start --host 127.0.0.1 --port 9000
oneopen start --reload
Option |
Default |
Description |
|---|---|---|
|
|
Bind address |
|
|
Bind port |
|
off |
Auto-reload (development) |
ASGI target: oneopen_ml_studio.server.app:app.
oneopen serve
Alias for start.
oneopen init
oneopen init
oneopen init --data-dir D:\oneopen-data
Creates the data directory, SQLite DB, and standard folders.
oneopen version
oneopen version
# OneOpen ML Studio v1.0.0
oneopen user create-admin
Create or update a platform admin (useful for Team/Enterprise bootstrap).
oneopen user create-admin --email admin@example.com --password 'strong-password' --name Admin
oneopen worker
Start a Celery worker (requires ONEOPEN_REDIS_URL and the [server] extra).
pip install "oneopen-ml-studio[server]"
export ONEOPEN_REDIS_URL=redis://localhost:6379/0
oneopen worker --concurrency 1
Option |
Default |
Description |
|---|---|---|
|
|
Worker processes (keep low for GPU training) |
|
|
Celery log level |
Future CLI (roadmap)
Additional commands such as project create, dataset import, dataset validate, train start, experiment list, and model export may land as the registry and adapter layers mature. Most of those flows are available today via the UI and HTTP API.