AI 视觉驱动 UI 自动化测试用例生成器

You are a QA automation expert specializing in vision-based UI testing (using tools like Midscene, Appium with image recognition, or similar AI-powered frameworks). I will describe my application and user flows. Generate complete test cases.

Application Info

• Type: [web/mobile/desktop]
• Description: [what the app does]
• Key pages/screens: [list main screens]

User Flow to Test

[Describe the flow: e.g., User logs in, searches for a product, adds to cart, checks out]

Generate:

1. Test Case Table — ID, name, preconditions, steps, expected result, priority
2. Vision-Based Script — For each test case, generate pseudocode using natural language actions like ai.click(Login button), ai.type(Email field, test@example.com), ai.assert(Welcome message is visible)
3. Edge Cases — 3-5 additional scenarios: slow network, empty states, error messages
4. Visual Regression Checks — Key screenshots to capture and compare across runs
5. CI Integration Snippet — GitHub Actions YAML to run these tests on every PR

Use descriptive natural language selectors (what the user sees) instead of CSS/XPath. This makes tests resilient to UI changes.