Documentation
Railwarden runs Ansible, Terraform, OpenTofu, Bash, PowerShell, Python, and Go across a fleet and treats every run as
structured data. One binary: serve is the API, the executor, the scheduler, and the web UI.
worker adds capacity.
State lives in one database, SQLite by default or PostgreSQL by DSN. These pages also render inside
the app at /ui/docs.
| Guide | What |
|---|---|
| Quickstart | Zero to a first run in a few minutes. |
| Switching from AWX | Import what you have, or set up from scratch. |
| Tutorials | Task-focused walk-throughs for everyday work. |
| Concepts | Runs, splits, pipelines, projects, templates, and the rest. |
| Reliability | How runs execute: workers, splits, failure, recovery, durability. |
| Configuration | Every command, flag, and environment variable. |
| Desktop | Run Railwarden as a local desktop app. |
| Features | The full capability list. |
| Advisory AI | The five AI features, the guarantees, providers, and what a model sees. |
| Extend in Go | The SDK: add tools, AI providers, secret engines, and notifiers. |
| HTTP API | Every endpoint the server exposes. |
| Migration | Moving off AWX or Semaphore in detail. |
| Comparison | How Railwarden compares to AWX and Semaphore. |
For deployment, the repository root holds a docker-compose.yml for a server, a database, and a
worker, and deploy/helm holds a Helm chart.