Dagu — Local-First DAG Workflow Engine with Web UI

Introduction

Dagu is a lightweight, local-first workflow engine that uses directed acyclic graphs (DAGs) defined in YAML to orchestrate tasks. It provides a built-in web UI for managing, monitoring, and debugging workflows without requiring an external database or message broker — making it ideal for small teams and individual developers who need cron-like automation with visibility.

What Dagu Does

• Executes multi-step workflows defined as DAGs in declarative YAML files
• Provides a web-based dashboard for scheduling, monitoring, and manual triggering
• Handles step dependencies, retries, conditional execution, and parallel steps
• Sends notifications on success or failure via email, Slack, or webhook
• Logs execution history with stdout/stderr capture for each step

Architecture Overview

Dagu is a single Go binary with no external dependencies. It reads DAG definitions from a local directory, schedules them based on cron expressions, and executes steps as subprocesses. Execution state and history are stored as local JSON files, eliminating the need for a database. The embedded web server provides the management UI using server-rendered HTML.

Self-Hosting & Configuration

• Single binary deployment with no database, Redis, or message queue required
• DAG definitions are YAML files stored in a configurable directory
• Schedule workflows using standard cron expressions within the YAML
• Configure global settings via a single config file or environment variables
• Runs on Linux, macOS, and in Docker containers

Key Features

• Zero-dependency single binary with embedded web UI
• YAML-based DAG definitions with step dependencies and parallelism
• Built-in scheduler with cron expression support
• Execution history with log viewing and manual retry from the UI
• Parameterized workflows with environment variable passing between steps

Comparison with Similar Tools

• Apache Airflow — Python-based with heavy infrastructure needs; Dagu is a single binary
• Cron — no dependency graph or UI; Dagu adds DAG execution and a dashboard
• Temporal — distributed workflow platform; Dagu targets local-first simplicity
• n8n — visual node-based automation; Dagu uses code-first YAML definitions

FAQ

Q: Does Dagu require a database? A: No. Dagu stores all state as local files. No PostgreSQL, MySQL, or Redis needed.

Q: Can Dagu run Docker containers as steps? A: Yes. Steps can execute any command, including docker run, making it easy to run containerized tasks.

Q: How does it handle failures? A: You can configure per-step retry policies, and the UI lets you manually retry failed steps or restart entire DAGs from a specific point.

Q: Is there an API for programmatic access? A: Yes. Dagu exposes a REST API for triggering, stopping, and querying workflow status.

Sources

• https://github.com/dagucloud/dagu
• https://dagu.readthedocs.io