Dagu

Dagu is a workflow orchestration tool built for teams whose core work isn't orchestration itself. It runs as a single binary with no external database, no framework to manage, and no infrastructure to provision before you can schedule your first job. Workflows are defined in declarative YAML and can run locally, over SSH, or inside containers, without touching your existing scripts.

It's a practical alternative to heavier tools like Airflow when you need reliable scheduling and visibility but don't want to operate a separate data platform to get there. State is stored in local files, which means there's nothing to size, migrate, or scale separately.

Key capabilities:

• Cron scheduling with timezone support, catchup windows, overlap policies, and per-step retry policies
• Web UI with real-time monitoring, a visual DAG view, and execution history
• Distributed execution via a coordinator that dispatches tasks over gRPC to workers selected by labels (e.g., gpu=true)
• Multiple executors including Docker, SSH, and HTTP out of the box
• Human tasks for approval steps or manual interventions inside automated workflows
• MCP integration so you can build and trigger workflows directly from AI coding tools like Claude, Cursor, or Copilot
• Notifications via email and webhooks, with lifecycle handlers and zombie detection
• Secret masking and per-DAG webhook tokens for secure automation

The same binary covers three deployment shapes: a single all-in-one server, temporary per-run workers provisioned by your platform, or a coordinator-and-workers setup for larger environments.

The community edition is free under GPLv3 and covers unlimited servers and workers. Paid licenses add SSO, RBAC, audit logging, incident routing, and priority support.