What is Dagu?

Dagu is a powerful, self-contained workflow orchestration engine that runs as a single binary with zero external dependencies—no database required. Despite its lightweight design, Dagu includes robust features such as built-in queuing, logging, execution history, and fully featured web UI for visualization. It’s designed to eliminate the complexity of workflow management without the overhead of traditional orchestration platforms.

Why Dagu Exists

Picture this: hundreds of cron jobs scattered across multiple servers, written in various languages, with hidden dependencies that only a few people understand. When something breaks at 3 AM, you're SSHing into servers, hunting through logs, trying to piece together what went wrong.

Sound familiar? That's exactly why Dagu was created.

Many organizations still rely on these legacy job scheduling systems. The scripts might be in Perl, Shell, Python, or a mix of everything. Dependencies are implicit—Job B assumes Job A created a certain file, but this isn't documented anywhere. Recovery requires tribal knowledge that disappears when team members leave.

Dagu eliminates this complexity by providing a clear, visual, and understandable way to define workflows and manage dependencies.

Vision & Mission

Dagu’s vision is to make software operations - including AI-agents - distributed and autonomous. Instead of relying on large, centralized orchestration platforms, Dagu enables scalable and simple operations by allowing multiple software components and AI agents to work together asynchronously, using the file system as a common hub.

Our core principles:

• Free from language lock-in – Use any tool, any language
• Runs anywhere – From your laptop, or to production servers
• Minimal overhead – Single binary, no additional infrastructure needed
• Accessible to all – Clear YAML syntax that anyone can understand
• Self-contained – Each process operates independently, needing only the file system

We're stripping away unnecessary complexity to make robust workflows accessible to everyone, not just specialized engineers.

Core Principles

Dagu was born from a simple observation: existing workflow tools either lack features (like cron) or require too much commitment (like Airflow, Prefect, or Temporal forcing you into Python or Go ecosystems).

We built Dagu around six core principles:

1. Local First

Define and execute workflows in a single, self-contained environment—no internet required. Whether you're prototyping on your laptop, running on IoT devices, or deploying to air-gapped on-premise servers, Dagu just works.

2. Minimal Configuration

Start with just:

• One binary
• One YAML file
• That's it!

No external databases. No message queues. Local file storage handles everything—DAG definitions, logs, and metadata. Complex infrastructure shouldn't be a prerequisite for workflow automation.

3. Language Agnostic

Your workflows, your choice:

steps:
  - name: python-task
    command: python analyze.py
  - name: bash-task  
    command: ./process.sh
  - name: docker-task
    executor:
      type: docker
      config:
        image: node:18
    command: npm run build

Any runtime works without extra layers of complexity. Use the tools your team already knows.

4. Keep it Simple

Workflows are defined in clear, human-readable YAML:

steps:
  - name: download data
    command: curl https://api.example.com/data.json -o data.json
    
  - name: process data
    command: python process.py data.json
    depends: download data
    
  - name: upload results
    command: aws s3 cp results.csv s3://my-bucket/
    depends: process data

Simple to understand, even for non-developers. Fast onboarding for new team members.

5. Community-Driven

As an open-source project, Dagu evolves with its users:

• Contribute new executors
• Integrate emerging technologies
• Propose enhancements via GitHub
• Share workflows and best practices

Real-world use cases drive development, keeping Dagu practical and aligned with what teams actually need.

Key Features

Zero Dependencies

One binary. Works on Linux, macOS, Windows. No database, no message broker, no runtime dependencies.

Hierarchical DAG Composition

Build complex workflows from simple, reusable components:

steps:
  - name: data-pipeline
    run: etl.yaml
    params: "SOURCE=prod"
  - name: ml-training
    run: train.yaml
    depends: data-pipeline

Built-in Web UI

Monitor workflows, view logs, and manage executions through a clean, modern interface. No additional setup required.

Production Ready

• Robust error handling with configurable retries
• Comprehensive logging with stdout/stderr separation
• Graceful shutdown and cleanup
• Health checks and metrics for monitoring
• Signal handling
• Notifications via email or webhooks

When to Use Dagu

Dagu is perfect for:

• Data Engineering: ETL pipelines, data processing, batch jobs
• DevOps Automation: Deployments, backups, system maintenance
• Local Development: Test workflows on your laptop before deploying
• Legacy System Modernization: Replace scattered cron jobs with managed workflows
• IoT & Edge Computing: Run workflows on resource-constrained devices
• AI agents: Orchestrate AI workflows with clear dependencies and execution history

When NOT to Use Dagu

Dagu might not be the best choice for:

• Workflows requiring sub-second scheduling precision
• Real-time streaming data processing (consider Apache Flink or Spark)

Continue to Getting Started →

Or explore examples to see what Dagu can do.