Dagu Blog
Essays and practical notes on command workflows, automation, and operating Dagu in production.
Dagu's redesigned web UI: one page to edit and run a workflow
A workflow in Dagu now has one page. Edit it on a canvas or as YAML, start it, and follow its runs, logs, and approvals in the same place. Saving still writes an ordinary YAML file.
Open, closed, inner, outer: the loops around a coding agent
Everyone runs coding agents in loops now, and everyone means something different by the word. Defining the four loop terms properly turns out to be the same thing as designing the loop.
Kitewell beta: Dagu as a desktop app, with an AI assistant
Kitewell, a macOS and Windows app built on Dagu, is in beta. Describe a job to its AI assistant, check the workflow on a visual map, and your computer runs it with the Dagu engine inside. It is free for one project.
How a Dagu server behind a firewall takes webhooks
Public webhook URLs reach Dagu servers that nothing on the internet can connect to. How the server holds the connection, why a request delivered twice starts one run, and why signatures are checked on the server against a time Dagu Cloud records.
Start Dagu workflows from GitHub, Stripe, and Slack, wherever Dagu runs
Dagu v2.20 gives a DAG's webhook a public URL through Dagu Cloud, so GitHub, Stripe, or Slack can start it on a server with no public address. Requests wait while the server is offline.
Monitor one Dagu server for free
Dagu Console now monitors one Dagu server per workspace for free: its health, its runs, and an email when a workflow fails or the server goes quiet. It needs Dagu v2.19.2.
Monitoring Dagu servers without seeing their logs
A hosted monitor for self-hosted servers has to answer one question before any other: what leaves the server? How Dagu's reporter decides, survives outages without losing or repeating a run, and tells silence from failure.
Introducing Agent DAGs: let the model choose what runs next
The industry is arguing about whether agents should run as loops or graphs. I think both miss the same thing: give the agent a task list and let it reconcile.
Inside Dagu v2.7.0: Reproducible Workflows
Dagu v2.7.0 makes workflow runs easier to reproduce by moving action versions, files, tools, inputs, outputs, and run evidence into the workflow contract.
The Gap Between Cron and Airflow
Cron is too small, Airflow is too big, and most teams live in the gap between them. Here is why that gap exists and what belongs in it.
Own Your AI Harness: Why Workflow Sovereignty Will Define the Next Decade
Claude and OpenAI are racing past 'just a model' into platforms and harnesses. That is a lock-in trap, and keeping control of your AI workflows matters.
Complete Guide to SubDAGs
A comprehensive guide to using SubDAGs in Dagu. Learn how to organize complex workflows by breaking them into reusable, modular components for better maintainability and scalability.
Using SSH Executor in Dagu: A Complete Guide
A comprehensive guide to using Dagu's SSH executor for running workflows on remote servers. Learn how to configure SSH authentication, manage remote tasks, and build distributed workflows.
Understanding Partial Success in DAG Systems: Building Resilient Workflows
Learn how to handle partial success scenarios in DAG-based workflow systems and build more resilient workflows with proper error handling strategies.
Dagu – L’orchestrateur de workflows pour oublier les crontabs
A French tech blog post exploring Dagu as the ideal middle ground between simple cron jobs and heavy orchestrators like Airflow. Covers how Dagu’s YAML-based DAG definitions bring clarity to complex job dependencies without requiring script rewrites.