Seven projects built through pull requests.

A game started from nothing. A developer tool and a paid product built from plans. A business platform of more than 770 pull requests. A developer web app and a ten-year-old mobile app that kept shipping. And the website you are reading. All of their changes went through ProPR as issues, branches and pull requests.

Four of the projects carry their names and links. Two are private, described with their owners' permission and with identifying details left out. The named projects and this site are built by the developer who builds ProPR, shown as examples of what the workflow produces. The counts come from GitHub as of 11 November 2026.

Overview

What the seven have in common.

A game, two products, a developer tool, a business platform, a decade-old app and a website share one path for their changes, and that path keeps AI-built software easy to steer whether the project is a week old or ten years old.

  1. Write down what you want.For a new project, a plan describes the product and ProPR turns it into small issues written in plain language. For an existing codebase, the backlog already holds the issues.
  2. One pull request per issue.An agent works on each issue on its own branch and sends back a pull request with the proposed change laid out for you to read.
  3. Review, then merge.Related pull requests collect in an epic, a branch for one feature. Nothing reaches the product until you merge, and GitHub keeps every merged pull request in the history with a Revert button.

Start your own.

ProPR Hosted runs everything for you; connect GitHub, describe what you want, and review the first pull request in your browser. Or try the live demo first.

See ProPR Hosted →

If you are new to GitHub, Your first AI pull request, start to finish explains every word, and self-hosting on your own server is free and open source.