Introduction

Learn what Rehearsal does, how it finds bugs and writes tests, and where to begin.

Rehearsal tests your apps like a real QA and finds the bugs before your users do. Its agents use your app in real browsers, the way your users do, and every bug they find comes with the steps and a recording of it happening. What they saw working becomes tests you run before each release, so you learn when a flow that used to work breaks.

It is for developers and teams who ship a web app and want it checked before every release without writing the tests by hand. Rehearsal starts from your app's address, and through its CLI it can also reach an app that runs on your computer.

Say your app is Parcel, at https://parcel.example, where people sign up, save cards, check out and invite their team. You give Rehearsal the address and ask it to check saved cards first. Re:agent, the assistant in the app, goes through Parcel's pages and writes a plan, and when you press Start testing, subagents work through the parts of the app, each in its own browser.

One subagent saves a card, signs out and signs back in, and the card is gone. A second subagent repeats those steps in a fresh browser and sees the same thing, so the bug goes into the report. This is how it reads when you copy it with Copy bug description:

Output
Bug #5: Saved card disappears after signing inSteps to reproduce1. Opened parcel.example/settings/cards2. Clicked Add card3. Filled Card number4. Clicked Save card5. Clicked Sign out6. Filled Email7. Filled Password8. Clicked Sign in9. Opened parcel.example/settings/cardsExpectedA saved card is still listed after the person signs out and signs in again.What happenedThe list of saved cards was empty after signing in again.Pagehttps://parcel.example/settings/cards

In the app, the same report sits beside the recording, with the bug's steps marked on it. Copy prompt for your coding agent copies a prompt that asks Claude Code, Codex or Cursor to fix it. Once the fix is in, Rehearsal can turn the bug into a test that fails if the card ever disappears again.

How it works Link to How it works

  • Explore. You tell Rehearsal what to check first, or to explore the whole app, and what to leave alone. Re:agent writes a plan of the parts it will test and waits until you press Start testing. Then subagents work through those parts while you watch their browsers, and a question for you, such as how to sign in, arrives as a message in the chat.
  • Bugs. A failed check usually becomes a bug only after a second subagent repeats it in a fresh browser and sees it fail again. Each bug holds what was expected, what happened, the steps and the recording, and you can hand it to your coding agent, mark it fixed or say it is not a bug. Small visual problems are reported as suggestions instead, and the pages are checked against WCAG 2.2 for a separate accessibility report.
  • Tests. When the exploration ends, Rehearsal recommends your main journeys, such as signing up or checking out, and one press turns what the subagents saw working there into tests, up to 20 at a time. Each test is a Retest test, and it is ready once it has passed twice, the second time in a fresh browser.
  • Runs. A run checks your ready tests at one address of your app, such as production or a preview. Every result keeps its steps and recording, and a failure shows the step where the test stopped and what the page showed. Exploring and making tests are free, and runs use test runs from your plan.

Ways to use Rehearsal Link to Ways to use Rehearsal

Everything above happens in the app at app.rehearsal.dev, in the chat with Re:agent and the panel beside it. The Rehearsal CLI lets Rehearsal's browsers reach an app on your computer, and runs your tests from a script or CI. Your coding agent, such as Claude Code, Codex or Cursor, can start an exploration, read the bugs, fix your code and ask Rehearsal to check each fix. And once you connect a GitHub repository and add a rule, your tests run on pull requests and their results appear as checks on the commit.

Get started Link to Get started