Logo
Deploy Now

SaaS Alternative

AWS Step Functions Apache Airflow Prefect Temporal

Categories

Self-Hosted DevOps CI/CD Workflow Automation AI Operations DAG Scheduler Task Orchestration

Stars

3,785

Forks

312

Watchers

22

Developer links

Last Updated: 08-21-2026 04:37 PM (UTC)

Dagu

With over 3,700 GitHub stars and growing adoption among teams tired of managing complex orchestration platforms, Dagu delivers a complete workflow engine as a single Go binary that requires no external database, message broker, or framework installation. Define directed acyclic graphs in declarative YAML — specifying dependencies, schedules, retries, timeouts, approvals, and notifications — while keeping existing shell scripts, containers, and tools completely unchanged. The built-in Web UI provides live pipeline visualization, per-step log streaming, run history browsing, artifact previewing, manual retry controls, and workflow YAML editing without SSHing into servers. Execute steps as shell commands, Docker containers, Kubernetes Jobs, SSH remote commands, SQL queries, or HTTP requests, with conditional branching and parallel execution handled natively. The integrated Model Context Protocol server exposes dagu_read, dagu_change, and dagu_execute tools, enabling AI agents like Claude, Codex, and Cursor to inspect workflow state, preview YAML modifications, and control runs through authenticated endpoints. The harness.run executor lets external coding-agent CLIs operate inside DAG steps with full scheduling and approval gate support. Scale beyond a single machine with the distributed worker mode, which dispatches tasks to remote nodes via gRPC with automatic label-based routing and worker selection. Deploy with Docker, the official Helm chart for Kubernetes, or a simple binary download requiring only a Linux, macOS, or Windows host. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPLv3 licensed.

Dagu
Dagu
Dagu
Dagu
Dagu

Benefits

  • Zero-Infrastructure Workflow Orchestration
  • Runs as a single Go binary with file-backed state storage, eliminating the need for PostgreSQL, Redis, RabbitMQ, or any external database or message broker dependency.
  • AI-Native Workflow Management
  • Built-in MCP server lets Claude, Codex, and Cursor inspect DAG state, preview YAML edits, and control workflow runs through authenticated REST and SSE endpoints.
  • Universal Step Executors
  • Natively runs shell commands, Docker containers, Kubernetes Jobs, SSH remote commands, SQL queries, HTTP requests, and external coding-agent CLIs without framework modifications.
  • Scales From Laptop to Fleet
  • Start with a single binary on one machine, then add distributed workers via gRPC with automatic label-based routing when workloads outgrow local execution capacity.

Features

  • Declarative YAML DAGs
  • Define workflow dependencies, parameters, schedules, retries, timeouts, and approval gates in human-readable YAML files versionable in Git.
  • Web UI Dashboard
  • Visualize pipeline graphs, stream per-step logs, browse run history, preview artifacts, and retry or stop workflows from the browser.
  • MCP Protocol Server
  • Built-in /mcp endpoint exposes dagu_read, dagu_change, and dagu_execute tools for AI agent integration with full audit attribution.
  • Distributed Workers
  • Scale execution across multiple nodes using gRPC-based coordinator with configurable worker selectors and label-based task routing.
  • Cron Scheduling
  • Built-in cron-compatible scheduler triggers workflows on time-based schedules with automatic retry policies and configurable timeout controls.