What Greenroom does
An AI virtual user that walks the screens your pull request changed and posts an evidence-backed QA handoff.
Greenroom is a virtual user for your app. On every pull request, it works out which screens your diff touches, walks them the way a person would (on a real iOS simulator or a real browser), and posts a QA handoff on the PR: what it tried, what it saw, and what broke, with screenshots and an action journal as evidence.
It is not a test suite you write. There are no scripts to record and no selectors to maintain. Greenroom reads your change, forms goals ("add an exercise to Monday's plan", "complete checkout with the promo applied"), and pursues them against the built app.
How a run fits your workflow
- You install the GitHub App on a repository. Greenroom's access is read-only, always.
- You add one workflow file and two config files to the repo.
- You open a pull request. CI builds your app without secrets, and Greenroom walks the changed screens.
- A QA handoff lands on the PR: verdicts per goal, defects with severity, evidence for every claim, and a pointer into your source for each finding.
Where to start
- Web quickstart if your app runs in a browser.
- iOS quickstart if your app builds for the iOS Simulator.
- How a walk works if you want the mental model first.
What makes the verdicts trustworthy
Every claim in a handoff is backed by captured evidence, and every run ends in an explicit verdict. When Greenroom cannot finish a goal, it does not guess or quietly stop: the goal fails closed as blocked, with the reason stated. When your app's declared state contract is missing, verdicts are capped at inconclusive rather than dressed up as passes. The verdicts page explains the vocabulary.