Lightweight and powerful workflow engine

Define workflows in YAML. Execute with a single binary. No database or message broker required. Ideal for VMs, containers, and bare metal.

Get StartedView Examples

Demo

CLI: Execute workflows from the command line.

Web UI: Monitor, control, and debug workflows visually.

Try It Live

Explore without installing: Live Demo

Credentials: demouser / demouser

Why Dagu?

• Single binary - No database, message broker, or external services. Architecture
• Declarative YAML - Define workflows without code. YAML Reference
• Composable - Nest sub-workflows with parameters. Control Flow
• Distributed - Route tasks to workers via labels. Distributed Execution
• Production-ready - Retries, hooks, metrics, RBAC. Error Handling

Quick Start

Install

macOS/Linux

curl -L https://raw.githubusercontent.com/dagu-org/dagu/main/scripts/installer.sh | bash

Windows (PowerShell)

irm https://raw.githubusercontent.com/dagu-org/dagu/main/scripts/installer.ps1 | iex

Windows (CMD)

curl -fsSL https://raw.githubusercontent.com/dagu-org/dagu/main/scripts/installer.cmd -o installer.cmd && .\installer.cmd && del installer.cmd

Docker

docker run --rm -v ~/.dagu:/var/lib/dagu -p 8080:8080 ghcr.io/dagu-org/dagu:latest dagu start-all

Homebrew

brew install dagu

npm

npm install -g --ignore-scripts=false @dagu-org/dagu

Create a Workflow

cat > hello.yaml << 'EOF'
steps:
  - command: echo "Hello from Dagu!"
  - command: echo "Step 2"
EOF

Run

dagu start hello.yaml

Start Web UI

dagu start-all

Visit http://localhost:8080

Key Capabilities

Capability	Description
Nested Workflows	Reusable sub-DAGs with full execution lineage tracking
Distributed Execution	Label-based worker routing with automatic service discovery
Error Handling	Exponential backoff retries, lifecycle hooks, continue-on-failure
Step Types	Shell, Docker, SSH, HTTP, JQ, Mail, and more
Observability	Live logs, Gantt charts, Prometheus metrics, OpenTelemetry
Security	Built-in RBAC with admin, manager, operator, and viewer roles

Example

A data pipeline with scheduling, parallel execution, sub-workflows, and retry logic:

schedule: "0 2 * * *"
type: graph

steps:
  - name: extract
    command: python extract.py --date=${DATE}
    output: RAW_DATA

  - name: transform
    call: transform-workflow
    params: "INPUT=${RAW_DATA}"
    depends: extract
    parallel:
      items: [customers, orders, products]

  - name: load
    command: python load.py
    depends: transform
    retryPolicy:
      limit: 3
      intervalSec: 10

handlerOn:
  success:
    command: notify.sh "Pipeline succeeded"
  failure:
    command: alert.sh "Pipeline failed"

See Examples for more patterns.

Use Cases

• Data Pipelines - ETL/ELT with complex dependencies and parallel processing
• ML Workflows - GPU/CPU worker routing for training and inference
• Deployment Automation - Multi-environment rollouts with approval gates
• Legacy Migration - Wrap existing scripts without rewriting them

Quick Links: Overview | CLI | Web UI | API | Architecture

Learn More

Overview

What is Dagu and how it works

Getting Started

Installation and first workflow

Writing Workflows

Complete workflow authoring guide

YAML Reference

All configuration options

Features

Scheduling, queues, distributed execution

Configuration

Server, authentication, operations

Community

GitHubDiscordBlueskyIssues