Quick start

Go from a new account to a first exploration of your app, its bugs and tests you can run again.

This guide takes you from a new account to a first exploration of your app, the bugs it finds and a first run of tests made from it. You need your web app's address. If the app signs people in, keep an account ready for Rehearsal to use, because a subagent may ask you for one.

An app on your computer

If your app runs only on your computer, enter its local address, such as localhost:3000. Setup then shows the commands that install the Rehearsal CLI and connect the app, and goes on once the connection is open. The CLI needs Node.js 24.12 or newer, and Test an app on your computer explains the connection.

Create your account Link to Create your account

  1. Step 1: Sign up

    Go to app.rehearsal.dev and choose Sign up with Google or Sign up with email. With email, Rehearsal sends you a link that confirms the address.

  2. Step 2: Name yourself and your workspace

    Setup asks for your name, then for a workspace, the space you share with your team. If you signed up with a work address, the workspace is named after your company, and its address stands under the name, such as app.rehearsal.dev/parcel. Press Change if you want another address.

    Welcome to Rehearsal comes next and shows what the agents do. Press Continue.

Tell Rehearsal what to explore Link to Tell Rehearsal what to explore

The next screens ask one question each. Your answers become the brief Re:agent works from, so the chat does not ask them again.

  1. Step 1: Enter your app's address

    If setup first asks What are you testing?, choose Web app. Then type your app's address, such as parcel.example. Rehearsal opens the page from its own servers, and once the line under the field says Opened with your app's name, press Continue.

    If the page is a sign-in page, setup says so, and Re:agent asks you for a test account in the chat later. If the page links to your company's other sites, such as docs at docs.parcel.example, setup asks Which one should we test? before going on.

  2. Step 2: Choose what to check first

    What should we check first? offers Signing up and signing in, Getting started after signing up and The main features, which you can combine, or The whole app on its own. The last row takes your own words, such as "Saved cards". Choosing one of the first two also allows the exploration to create an account in your app.

  3. Step 3: Choose what to leave alone

    What should we leave alone? offers Settings and account pages and Checkout and payments, or Nothing, go ahead, and again a row for your own words. Whatever you choose here, an exploration never pays for anything.

  4. Step 4: Start exploring

    The last screen, Ready to explore Parcel?, shows what Re:agent is told: your app's address, what to check first and what to leave alone. Press Start exploring, and the chat opens on the exploration.

Approve the plan Link to Approve the plan

Re:agent first goes through your app on its own. The chat says Going through the app and planning and counts the pages it opens, with pictures of the newest ones. If it needs to sign in to go further, it asks how to get in, as a question above the message box: choose a test account, add one there with Add a test account, or press Go on without an account.

Then the plan appears in the chat as one card, Exploration plan. It has a row for each part of your app, such as Sign up, Saved cards and Checkout, with how many checks it holds, and pressing a row shows those checks in the order the subagents will follow them. A part's switch turns it off. You can also type a change in the message box, such as "Also test inviting a teammate", and Re:agent changes the plan. The foot of the card says about how long the work will take.

The subagents start only when you press Start testing, and the plan waits for that however long you take.

Before you press Start testing

The subagents use your app for real, so whatever they add, such as a saved card, is really added. What the plan adds to your site lists those things, and What the plan won't do lists what they never do, such as paying for anything, deleting anything they did not make or messaging real people. If nothing should be added to production, give Rehearsal a staging address instead.

Watch the subagents Link to Watch the subagents

Once you press Start testing, subagents start on the parts that are on, each with a browser of its own on Rehearsal's servers. The chat names each one as it starts, and the panel beside the chat follows them all.

  • Subagents lists every subagent with the last step it took. Open one to watch its browser as it works, with each step listed under the picture.
  • Test cases lists what the subagents check, one sentence per case, grouped by part of your app and then by page. Each case gets a mark once it is checked, such as passed, a bug, or Blocked when the subagent could not check it.
  • Recordings keeps the recording of every browser the subagents used.

A subagent that needs something only you know, such as which account to sign in with, asks in the chat, and its question waits above the message box. Answer while the exploration still has time, because five minutes after its time runs out, the subagents still on its parts are stopped, waiting or not. Questions from Re:agent explains the answers you can give. To end the exploration early, press Stop exploring in the message box, and you still get the report for what was checked.

Read your first bug Link to Read your first bug

When a check fails, a second subagent usually repeats it in a fresh browser. If it fails again, the bug appears under Bugs in the panel with its number, such as Bug #5, and while the sidebar is open, the number of open bugs shows beside Bugs there. A failure that nobody saw a second time is listed further down, under Bugs we couldn't reproduce.

Press a bug to open its report. It holds the recording with the bug's steps marked on it, then Steps to reproduce, Expected and What happened. Under them, Copy prompt for your coding agent copies a prompt you paste into Claude Code, Codex or Cursor to fix it, and Mark as fixed and Mark as not a bug record what you decided. Fix a bug and check the fix follows a bug from there.

Create tests from the main journeys Link to Create tests from the main journeys

When the exploration ends, the chat says Exploration complete. Here's the report. The report is one card with the test cases checked, the bugs found, the accessibility report and the suggestions. Under Test report, press Review the test cases.

The panel opens on a card that recommends your main journeys, such as signing up and checking out, with how many tests each would make. Press Create tests from the main journeys, or Choose parts yourself to tick the tests you want. Rehearsal makes up to 20 tests at a time, and making them is free.

Each test is a Retest test written from what a subagent saw work. Rehearsal runs every new test twice, the second time in a fresh browser, and the card counts a test as ready once both runs pass. A test that did not pass says Could not verify, and Retry tries again.

Run your tests Link to Run your tests

When every test has been checked, press Run once now on the same card to run the ready ones. The run dialog says what the run will check, such as "This run checks 5 tests on 1 screen: 5 test runs.", and lets you choose the screens and, if your app has more than one address, the environment to run on. Each screen you choose becomes a run of its own. Press Run.

Your tests now live under Tests in the sidebar, where Run test runs one of them, and ticking several and pressing Run selected runs them together. Every run is listed under Runs. Its page shows how each test ended, with the recording and the steps, and for a failure the step where the test stopped and what the page showed, as Read a run explains.

Runs of your tests use test runs from your plan, while exploring and making tests use none. Your workspace's first test run also starts its free week, when test runs are unlimited.

Next steps Link to Next steps

Once your first exploration has ended, Next steps stands at the foot of the chats list and names the one thing to do next, such as fixing the bugs with your coding agent or connecting GitHub. To explore again later, start a new chat in Re:agent and press Explore Parcel again, or type /explore.