---
title: "Introducing Orbit"
description: "Why we built Orbit: deploys that start as a readable plan and end with an instant undo."
date: "2026-05-12"
tags: ["announcement"]
canonical: "https://orbit-handbook.sitesmojo.com/blog/introducing-orbit/"
---

Today we are
putting Orbit's handbook and this blog on the public internet.

Orbit started as a shell script named `ship.sh` that three of us were too
embarrassed to show anyone. It rebuilt images, copied files, restarted
services, and worked right up until the night it did not - a half-finished
deploy at 11pm, one service on the new build, two on the old, and no record
of what `ship.sh` had actually done before it died.

The tool we wanted after that night had three properties:

1. **Show the change before making it.** A deploy should start as a plan a
   human can read.
2. **Make the same change everywhere.** Laptop and CI should run the same
   code path, not a script and its "temporary" fork.
3. **Make undo trivial.** If rolling back takes longer than deciding to,
   nobody rolls back - they patch forward at midnight instead.

Orbit is those three properties as a single binary. Describe your service in
[one file](/docs/configuration/), and `orbit deploy` plans, diffs, applies,
verifies, and keeps every release addressable for
[instant rollback](/docs/deployments/rollbacks/).

Start with the [five-minute quickstart](/docs/getting-started/). The
[roadmap](/docs/roadmap/) says where this goes next - previews per branch are
closest.

*Orbit is a fictional product; this site is a live demo of Mojo Sites - the
handbook and this blog are docit books mounted on a static site.*