🛠 CLI Reference

🛠️ CLI Quick Start

Go from zero to a running pipeline in 5 minutes using the FlowyML command line.

⚡ 5 Minutes 🌱 Init ▶️ Run 🖥️ Visualize

Full Reference

This page is a quick-start tutorial. For the complete list of every command, flag, and environment variable, see the CLI Reference.

Step 1 — Install FlowyML

pip install flowyml

Verify it worked:

flowyml --version

Step 2 — Create a Project

flowyml init my-project
cd my-project

This scaffolds everything you need:

my-project/
├── flowyml.yaml         # Stack & resource config
├── requirements.txt     # Python dependencies
└── src/
    └── pipeline.py      # Your first pipeline

Templates

Use --template ml or --template cv for pre-built ML or Computer Vision project structures.

Step 3 — Run Your Pipeline

flowyml run src/pipeline.py

You should see:

Fetching data...
Processing 5 items...
✓ Pipeline finished successfully!

That's it!

Your pipeline ran locally with caching enabled by default. Run it again — cached steps are skipped automatically.

Step 4 — Launch the UI

flowyml ui start

Open http://localhost:8080 to see the DAG visualization, run history, and artifact inspector.

Stop it when you're done:

flowyml ui stop

Step 5 — Serve or Deploy a Model

Once a pipeline has registered a model, serve it locally (no Docker) or deploy it to a target:

# In-process FastAPI server for quick checks
flowyml serve my_model --stage production

# Package + deploy to local Docker / Kubernetes / OpenShift
flowyml deploy my_model --stage production --runtime fastapi --target local_docker

# Inspect and call recorded deployments
flowyml deployment list
flowyml deployment predict my_model-endpoint --json '{"inputs": [[0.1, 0.9, 1.2]]}'

See the Model Serving & Deployment guide for runtimes, targets, promotion, and batch inference.

Step 6 — Pass Parameters

Override context values from the command line without changing code:

flowyml run src/pipeline.py \
  --context dataset_size=100 \
  --context multiplier=5

Step 7 — Try a Dry Run

See what would happen without actually executing:

flowyml run src/pipeline.py --stack production --dry-run

Environment Variables

All FlowyML environment variables use the uppercase FLOWYML_ prefix:

Variable	Default	Description
`FLOWYML_HOME`	`~/.flowyml`	FlowyML home directory
`FLOWYML_ENV`	`dev`	Environment name (`dev`, `staging`, `prod`)
`FLOWYML_UI_PORT`	`8080`	Default UI port
`FLOWYML_LOG_LEVEL`	`INFO`	Logging level

Handy Aliases

Add to your ~/.zshrc or ~/.bashrc:

alias fml='flowyml'
alias fml-run='flowyml run'
alias fml-ui='flowyml ui start'

🚀 What's Next?

📖 Full CLI Reference

Every command, flag, and environment variable in one place.

CLI Reference →

⚙️ Configuration

Deep dive into flowyml.yaml, stacks, and resource presets.

Configuration →

🎢 Core Concepts

Learn Pipelines, Steps, Context, and Assets in depth.

Core Concepts →