Quick Start

Boot an engine, sign a browser in, and run a real workflow — about five minutes end to end.

Step 1: Boot the engine

Start the engine with the web console. The --console flag serves the console and auto-enables HTTP:

sudo dxflow boot up --console

The engine reports where it answers. The Console address is the one to open in a browser:

Engine successfully booted (use 'CTRL+C' to stop)
● Server
  Socket       ~/.dxflow/unix.sock
  HTTP         http://127.0.0.1:80  (bound to 0.0.0.0)
● Console
  Address      http://127.0.0.1:80  (open in a browser)

This runs in the foreground and holds the terminal. Other boot modes:

sudo dxflow boot up                      # Unix socket only, no web server
sudo dxflow boot up --https --console    # serve over HTTPS
sudo dxflow boot up --console --daemon   # background service (Linux, SystemD or OpenRC)

See Advanced Boot Configuration for every flag and deployment pattern.

Step 2: Sign in to the console

Open http://localhost (or http://<your-server-ip>) in a browser. The engine authenticates every connection that arrives over HTTP, so the console opens on a sign-in dialog.

The quickest way in is a pairing code. Leave the engine running and open a second terminal on the engine host:

sudo dxflow engine pair

It prints a single-use code and waits:

● Pairing
  Permissions  SHELL, ARTIFACT, RUNTIME, WORKFLOW, AGENT
  Code         K7QM3XPD
  Waiting 30s for a console to use the code

Choose Pairing code in the browser and enter it. The terminal then shows who is asking and waits for your answer:

● Pairing request
  Address      203.0.113.9
  Agent        Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7)
  Approve? [y/N] y

Answer y and the console is signed in.

Step 3: Run your first workflow

The dxflow hub is a catalog of ready-to-run workflows you pull by name. Deploy JupyterLab:

dxflow workflow create hub://jupyter --identity notebook
dxflow workflow start notebook --override env.app.PASSWORD=my-strong-pass

Workflow created successfully: with 'notebook' identity and 'created' status
Workflow started successfully: with 'notebook' identity

Open http://localhost:8888 and sign in with the password you set. --override tunes a single run — here it replaces the PASSWORD variable on the app step, which otherwise defaults to dxflow.

Browse the rest of the catalog from the CLI:

dxflow workflow hub search genomics
dxflow workflow hub inspect fastqc

Step 4: Watch it run

dxflow workflow list              # identity, name, status, tags, link, created
dxflow workflow steps notebook    # per-step status and exit codes
dxflow workflow logs notebook --live   # stream stdout and stderr
dxflow workflow stats notebook    # live CPU and memory, aggregated across steps

Stop it when you're done, and remove it to reclaim the disk:

dxflow workflow stop notebook
dxflow workflow remove notebook

Step 5: Write your own workflow

A workflow is a YAML definition with a name and a list of steps. Each step names a container image and the runtime it runs on:

# hello.yml
name: hello-world
tags:
    - getting-started
steps:
    - name: hello
      runtime: docker
      image: alpine:latest
      command:
          - echo
          - "Hello from dxflow"

Create and start it in one call:

dxflow workflow create ./hello.yml --identity hello --start
dxflow workflow logs hello

Steps also take ports, volumes, env, and resources (CPU, memory, GPU), and run in phases so you can mix sequential and parallel work. A step defaults to sequential mode, which requires a command; use mode: parallel for long-running services that keep their image's own entrypoint.

Next steps

• User Interface — manage workflows, artifacts, and shells from the console
• CLI Reference — every command, including artifact, shell, key, and agent
• API Documentation — integrate dxflow into your applications
• Advanced Topics — boot modes, authentication, licensing, and remoting

Troubleshooting

For more, see the FAQ and Troubleshooting.