Dagu

Dagu is a self-hosted workflow engine for defining DAG workflows in YAML and managing runs, logs, and retries through a built-in web interface. It runs as a single binary locally or across distributed workers, without requiring an external database or message broker. Workflows can execute shell scripts, Docker containers, SSH commands, SQL queries, HTTP requests, chat completion steps, and Agent Harness steps. Scheduling includes cron with timezones, catchup windows, overlap and retry policies, zombie detection, lifecycle handlers, and notifications. Built-in actions cover HTTP, mail, SSH, SFTP, Git, S3, Redis, LLM, browser, computer, SQL, PostgreSQL, SQLite, and DuckDB. Dagu also supports MCP connections from clients including Claude Code, Codex, Cursor, VS Code, Gemini CLI, OpenCode, Zed, and Cline. It is open source under GNU GPL v3.0. The free Community plan includes unlimited servers and workers, with up to two API keys. Team costs $50 a month and Pro costs $150 a month; both add SSO, role-based access, audit logging, incident routing, and standard support.

Dagu provides a self-hosted workflow engine with scheduling, integrations, and run management. The free Community plan may suit individual use, while Team and Pro add organizational controls at a monthly cost.

Dagu suits users who want to define and operate automated workflows on self-hosted systems. Teams needing SSO, access controls, audit logging, or incident routing can consider its paid licenses.

Dagu is a self-hosted workflow engine for people who want YAML-defined DAGs and browser-based control of every run. It suits engineers coordinating scripts, services, and data tasks across their own infrastructure. Its strongest case is broad execution and scheduling without a database dependency; its main compromise is that team administration and the free plan's API-key allowance are limited.

Overview

Dagu packages workflow definitions, run history, logs, and retry handling in a built-in web UI. It runs as one binary on a local machine or across distributed workers, so a deployment does not depend on an external database or message broker. That lean architecture is practical for self-hosted installations where operational simplicity matters, while the GNU GPL v3.0 license keeps the core software open source.

The Community edition allows unlimited servers and workers, but only two API keys. Teams needing identity integration, granular permissions, or formal auditing must move to a paid self-host license. Dagu fits the broader Workload Automation Software category.

Key features

YAML DAGs can launch existing shell scripts, Docker containers, SSH commands, SQL queries, HTTP requests, chat-completion steps, and Agent Harness steps. This mix favors organizations that already have useful commands and services and want to connect them rather than rebuild tasks around a single runtime.

Cron scheduling includes time zones, catchup windows, overlap policies, and retry policies. Zombie detection, lifecycle handlers, notifications, workflow dependencies, and event triggers provide controls for missed, failed, or stalled work instead of treating each run as an isolated job. Cross-platform execution broadens where workers can operate.

Integration actions cover HTTP, mail, SSH, SFTP, Git, S3, Redis, LLM, browser, computer, SQL, PostgreSQL, SQLite, and DuckDB. MCP connections are supported from Claude Code, Codex, Cursor, VS Code, Gemini CLI, OpenCode, Zed, and Cline, giving AI-oriented clients a route into Dagu workflows.

Pricing

Community: free, billed free forever. It includes unlimited servers and unlimited workers, with up to two API keys. This is the sensible starting point for individual operators or small deployments that can manage access without SSO or role separation.

Team: $50 a month, billed at $50 per month or $500 per year (save two months). The license covers three self-host Dagu servers, unlimited workers, SSO through OIDC, RBAC, audit logging, incident routing, and standard support. It fits a small organization that needs centralized identity and accountability.

Pro: $150 a month, billed at $150 per month or $1,500 per year (save two months). It expands the allotment to 15 self-host servers while retaining the Team controls, making it appropriate for larger estates rather than a single installation.

Enterprise: custom pricing with a custom licensed-server allotment and priority support. A 14-day free self-host license trial requires no credit card. Team and Pro support is standard; community help comes through documentation, GitHub Issues, and Discord on a best-effort basis, with support handled by Dagu's maintainer.

Platforms

Dagu supports API, Linux, macOS, self-hosted, web, and Windows environments. Its self-hosted design lets organizations place servers and workers where their execution needs require, while the web interface provides a common management surface.

Who it's for

Dagu is best for developers and operations teams assembling heterogeneous workloads from scripts, containers, remote commands, databases, and web services. It is also a good fit when avoiding a separate database or broker is a priority. Individual users can get substantial capability from Community, but organizations that require SSO, RBAC, audit logs, or incident routing should budget for Team or Pro. Buyers seeking a hosted-first service or more than two API keys at no cost should look elsewhere.

Pros and cons

• Pro: Single-binary deployment avoids mandatory database and message-broker operations.
• Pro: Broad step and integration support connects established scripts, infrastructure, data stores, and AI clients.
• Pro: Time-zone-aware cron, catchup, overlap, retries, and zombie detection address real scheduling and failure conditions.
• Con: Community is limited to two API keys, which can constrain shared or automated access.
• Con: SSO, RBAC, audit logging, incident routing, and enhanced support are paid-only controls.

Alternatives

Choose ActiveBatch instead when you want a paid platform with a no-obligation quote and execution-agent points that vary by operating system. BMC Control-M is the alternative for buyers prepared for a $2,400-per-month, annually billed SaaS Starter Pack covering cloud and hybrid orchestration. Pick Argo Workflows for a free, open-source Kubernetes workflow engine. Kestra is another free option with a free trial and Apache 2.0 licensing. Apache Airflow suits users wanting a no-cost Apache-licensed project for API, Linux, macOS, self-hosted, and web environments. Flyte is a free platform with a trial for API, Linux, self-hosted, and web deployments. Luigi is a free Apache-2.0 Python package without a trial. Prefect is worth choosing when its free Hobby tier's two users, one workspace, five deployments, 500 serverless minutes per month, and seven-day retention match your needs.

Verdict

Choose Dagu if you want open-source, self-hosted orchestration that can run many kinds of existing tasks with strong scheduling and run controls, especially without adding a database or broker. Look elsewhere if two Community API keys are insufficient or if the administrative features you need must be included without a $50-per-month license.