🧠 AutoTest Agent – Generate Playwright Tests with AI

This project provides a powerful CLI tool that automatically generates Playwright Page Objects and test files using live website DOM analysis or natural language instructions — powered by OpenAI.

---

דף סיכום פרוייקט בעברית: https://docs.google.com/document/d/1tknbhK3KhRkFETfSod9D_vOvsPulytcThmFUAxalxnU/edit?usp=drive_link

סרטון הדרכה והדגמה: https://drive.google.com/file/d/1saxPoqzSDQZmDW6eTQgHw0lMNYm6rvvx/view?usp=sharing

📦 Features

• ✨ Auto-generate Page Object Model (POM) files from any URL
• 🧪 Create Playwright test suites using detected methods
• 💬 Support for natural language instructions (e.g. "Test the login flow with invalid passwords")
• ✅ Built-in beforeAll / afterAll hooks with browser setup
• ⚙️ Smart locator generation using stable strategies (data-testid, roles, placeholders, etc.)
• 🚀 Fully CLI-driven — no manual coding needed!

---

📁 Project Structure

/pages/                => Generated Page Object files  
/tests/                => Generated test files  
/src/                  => Source code of the generator  
curserRules.ts         => Custom locator & naming preferences  
index.ts               => Main CLI entry point  
promptBuilder.ts       => Prompt templates used for OpenAI requests  

---

🛠️ Installation

1. Clone the repository

git clone https://github.com/your-username/autotest-agent.git
cd autotest-agent

2. Install dependencies

npm install

3. Set your OpenAI API Key

You must provide your own OpenAI API key. The system uses it to generate prompts and code.

Create a .env file:

touch .env

Then add this line:

OPENAI_API_KEY=your-openai-key-here

Alternatively, you can export it:

export OPENAI_API_KEY=your-openai-key-here

💡 You can get your API key from: https://platform.openai.com/account/api-keys

---

🚀 Usage

Run the CLI tool with:

npm start

You'll be guided with the following:

1. Choose mode

   • DOM-based code generation from a live website
   • Natural language instruction ("I want to test the signup flow")

2. Choose or enter the URL

   • Use the default or enter a new target site.

3. (Optional) Provide test instructions in English
   (if you selected the natural language mode)

4. ✅ Sit back and watch as the system:

   • Scans the DOM
   • Creates a Page Object
   • Generates matching test.spec.ts files
   • Stores them in the correct folders

---

🧪 Generated Test Example

import { test, expect, Browser, BrowserContext, Page } from '@playwright/test';
import { LoginPage } from '../../pages/example.com/LoginPage';

test.describe.serial('LoginPage Tests', () => {
  let browser: Browser;
  let context: BrowserContext;
  let page: Page;
  let loginPage: LoginPage;

  test.beforeAll(async ({ browser: testBrowser }) => {
    browser = testBrowser;
    context = await browser.newContext();
    page = await context.newPage();
    loginPage = new LoginPage(page);
    await page.goto('https://example.com/login');
  });

  test.afterAll(async () => {
    await context.close();
  });

  test('should login with valid credentials', async () => {
    await loginPage.setEmail('user@example.com');
    await loginPage.setPassword('password123');
    await loginPage.clickLogin();
    expect(await page.url()).toContain('/dashboard');
  });
});

---

⚠️ Notes

• The generated locators prioritize semantic attributes (getByRole, getByPlaceholder, data-testid, etc.)
• The CLI creates new folders under /pages and /tests based on the target domain
• Natural language test generation supports English only (for now)
• Your OpenAI usage may incur costs depending on your token limits

---

🔮 Roadmap / Ideas

• Multi-page flows and navigation
• Support for other AI providers (Anthropic, Mistral)
• Add support for headless = true / CI mode
• Improve error handling and locator fallback strategies

---

🤝 Contributing

Pull requests are welcome! Please fork this repo and open a PR for review.

---

📄 License: menachem caspi caspi45@gmial.com

MIT