Skip to main content

CLI Commands

cdui is a thin launcher (cdui.cmd on Windows) placed at ~/.local/bin/cdui by the installer. If you haven't restarted your terminal yet, invoke the absolute path ~/CodefyUI/cdui start, or use python scripts/dev.py <cmd>dev.py re-execs into the venv's Python automatically.

Core commands

CommandDescription
cdui installInstall backend deps; download the prebuilt frontend (or build locally if pnpm is available).
cdui updateUpdate to the latest release (prebuilt path) or pull main (source build) and re-sync the frontend.
cdui startProduction mode — single uvicorn on :8000, in the background (no Node needed). --foreground / -f runs it attached.
cdui statusbtop / k9s-style dashboard: CPU, memory, disk, GPU, top processes, plus the server's PID and health. Refreshes live (every 2s; Ctrl+C to quit). Pass a number to set the interval (cdui status 1), or --once for a single frame.
cdui devDeveloper mode — backend :8000 + Vite HMR :5173 (requires Node + pnpm).
cdui buildBuild the frontend bundle locally (requires Node + pnpm).
cdui stopStop all services (including the background server).
cdui testRun backend tests.
cdui cleanRemove the virtualenv, node_modules, and frontend/dist.
cdui uninstallClean + remove the PATH launcher.

Plugin commands

CommandDescription
cdui plugin install <name|url>Install a plugin pack (catalog name like foundations, owner/repo[@ref], or a full GitHub URL).
cdui plugin listList installed plugin packs.
cdui plugin info <id>Show a pack's manifest, lessons covered, and node names.
cdui plugin search <query>Query the plugin catalog.
cdui plugin uninstall <id>Remove an installed plugin pack.

See Plugins for the full plugin workflow.

Background vs foreground

cdui start runs in the background by default — close the terminal and the server keeps running. Manage it with:

cdui status # live dashboard + health
cdui stop # stop the background server
cdui start -f # run attached instead (Ctrl+C to stop)

Running a graph without the server

You don't need the web UI to execute a graph — see CLI Graph Runner:

cd backend
python run_graph.py ../examples/Usage_Example/CNN-MNIST/TrainCNN-MNIST/graph.json