Dagu

The tool provides a self‑hosted control plane for existing operational automation, allowing teams to run, schedule, review, and debug tasks without requiring a database, message broker, or language‑specific SDK. Workflows are defined declaratively in YAML as Directed Acyclic Graphs and can execute shell commands, scripts, containers, HTTP requests, SQL queries, SSH commands, sub‑workflows, and AI‑agent steps. A single binary stores state in local files by default and adds features such as scheduling, dependencies, retries, queues, logs, and a Web UI, with optional distributed workers for scaling.

It is aimed at teams that already have scattered scripts, cron jobs, data pipelines, media‑processing jobs, or server‑automation tasks and want a unified view and governance for those operations. The system supports user management, RBAC, workspaces, approvals, secrets, a REST API, CLI, and webhooks, enabling collaborative operation and integration with existing processes.

Key distinctions include its local‑first design, the ability to orchestrate heterogeneous step types within a DAG, and the provision of a web‑based cockpit for monitoring run details, logs, and documents, all packaged as an open‑source, self‑hostable binary under GPL‑3.0.