# Getting started

Your first AI test, from a URL to a verdict, in a few minutes. Prefer to watch first? Here is the 6-minute quick start:

## 1\. Sign in to [Wopee.io Commander](<https://cmd.wopee.io>)

1. Go to [https://cmd.wopee.io](<https://cmd.wopee.io>).
2. Sign in with your [GitHub](<https://github.com/signup>), [GitLab](<https://gitlab.com/users/sign_up>) or [Google](<https://accounts.google.com/>) account.

You land on the **Projects** page. Every app you test lives here as one project.

## 2\. Create a project and start the first Analysis

1. Click **NEW PROJECT**.
2. Paste the URL of your app, or pick a demo (**E-shop demo**, **Website demo** or **Banking demo**). The demos come with test instructions already filled in.
3. Under **Test instructions**, describe in plain English what matters to you, including any test credentials. For example: *Log in with the test account, add an item to the cart, complete the purchase, and verify the "Thank you" message.* You can leave it empty and let the agent decide.
4. Click **Create project and generate tests**.

> [!TIP]
>
> **Advanced settings**
>
> The test framework (Playwright) is fixed once the project exists. Switch off **Explore the app and generate tests** to create an empty project and add your own tests first.

## 3\. Accept the GitHub invitation

Every project gets its own GitHub repository. To open it, accept the invitation: open the menu under your avatar, click **GH Invitation**, and click **Join** on GitHub.

## 4\. Watch the agent explore

The first Analysis starts right away. You see a live view of the agent's browser: it opens your app, logs in, clicks through the pages and checks what it sees. From what it saw, it writes:

- an **application description**,
- **user flows**, written as user stories,
- **test scenarios** for those flows, and runs every one of them, so you start with real results.

All of them are under **Variables &amp; Knowledge** and in the scenario list. Edit any of them, or add instructions and regenerate.

## 5\. Read the verdict

Click a scenario to open its run. The **Report** gives the verdict, every step with what was expected and what the agent verified, any issues it found, and a final analysis. When a test fails, read the report first: it tells you whether the app has a bug or the test needs a fix. A verdict is Passed, Failed or Incomplete; see [Verdict Integrity](<https://docs.wopee.io/concepts/verdict-integrity/index.md>).

Under **Steps**, every step has its Playwright locator and a screenshot. When your page changes, the agent can pick a better locator, so the test [heals itself](<https://docs.wopee.io/concepts/self-healing/index.md>). The **Playwright report** shows exactly what ran.

## 6\. Your test library

**Scenarios** lists every test, grouped by Analysis and user story, with its category, number of steps and the agent's last verdict.

To write your own test, click **\+ New test**, give it a name and describe it in plain English. On its first run the agent works out the steps; keep **Save steps from this run** switched on, and later runs replay them.

## 7\. Test Plans and schedules

Group tests into a **Test Plan** (for example *Smoke Test*), run it by hand with **Run**, or put it on a schedule: every hour, day, week, month, or a custom cron expression.

## 8\. Runs and visual checks

Every execution lands in **Runs**. Each step's screenshot is compared with its approved baseline, so you also see visual changes, not only broken clicks.

You can **👍 Approve** or **👎 Reject** the current state of the tested application if its [Test Step Status](<https://docs.wopee.io/glossary/#test-step-status>) is:

- `New` - no previous state was approved (for this test step)
- `Unresolved` - previous state was already approved but current state is different (for this test step)

## 9\. Your tests in your repository

Each Analysis has its own branch in the project's GitHub repository: the user flows are a Markdown file, and the scenarios, with every step and Playwright locator, are a JSON file. Every change is a commit. Open the repository from **More** › **Repo**, or the whole project in VS Code or Cursor.

> [!NOTE]
>
> **Free plan and step limits**
>
> The free plan gives the agent 50 steps every 5 hours and 2 projects. Every paid plan starts with the first month on us: see **Upgrade** in the app.

## More

- [🛠️ Tools &amp; Assertions](<https://docs.wopee.io/concepts/tools-and-assertions/index.md>): how the agent interacts with your app
- [🩹 Self-Healing](<https://docs.wopee.io/concepts/self-healing/index.md>): how tests keep working when your UI changes
- Run and generate tests from your coding agent (Claude Code, GitHub Copilot, Cursor, OpenCode and others): [🚀 Wopee.io MCP server](<https://docs.wopee.io/guides/wopee-mcp/index.md>)
- Integrate your existing tests: [Playwright](<https://docs.wopee.io/playwright-visual-testing/index.md>), [Cypress](<https://docs.wopee.io/cypress/01-getting-started/index.md>), [Robot Framework](<https://docs.wopee.io/robot-framework/01-getting-started/index.md>), [WebdriverIO](<https://docs.wopee.io/webdriverio-visual-testing/index.md>)
- Short videos: [Wopee.io in 2 minutes](<https://youtu.be/X4vKIrlauik>), [Visual checks](<https://youtu.be/jLA-rA1AHd0>), [Your tests, your repo](<https://youtu.be/7yLnLJBpBB4>)