To explore how AI can assist in QA workflows, I built a simple tool that:

• Analyzes a webpage
• Detects form elements
• Generates structured manual test cases automatically

The tool uses a local AI model and runs entirely on my laptop.

Problem Statement

When testing a webpage manually, a QA engineer typically performs the following steps:

1. Identify the input fields.
2. Identify available actions such as buttons.
3. Think of possible scenarios (valid, invalid, edge cases).
4. Write structured test cases.

For example, a login page normally requires tests like:

• Valid login
• Invalid password
• Empty fields
• Security scenarios

While these scenarios are straightforward, writing them repeatedly across multiple projects can become inefficient.

Idea

The idea behind this project was simple:

Let AI analyze the webpage structure and generate manual test cases automatically.

Workflow of the tool:

URL → DOM Analysis → AI Prompt → Generated Test Cases

Technologies Used

The project uses the following technologies:

• Streamlit – to create the web interface
• Beautiful Soup – to analyze webpage elements
• Ollama – to run AI models locally
• Phi-3 – to generate QA test cases

Running the AI model locally ensures that the tool works offline without sending data to external APIs.

How the Tool Works

Step 1 – Enter Website URL

The user enters the URL of the webpage to test.

Example:

https://the-internet.herokuapp.com/login

Step 2 – DOM Analysis

The tool scans the page and identifies important elements such as:

Inputs:

• username
• password

Buttons:

• Login

This is done using BeautifulSoup to parse the HTML structure.

Step 3 – AI Prompt Generation

The detected elements are sent to the AI model with a structured instruction like:

“Generate 5 manual QA test cases for a login page using the detected inputs and buttons.”

The prompt ensures the response follows a specific table format.

Step 4 – AI Generates Test Cases

The AI returns structured manual test cases such as:

Test Case ID | Scenario | Steps | Expected Result

TC_LOGIN_01 | Valid login | Enter valid username and password and click login | User successfully logs in TC_LOGIN_02 | Invalid password | Enter valid username and incorrect password | Error message displayed TC_LOGIN_03 | Empty fields | Leave username and password empty | Validation message displayed TC_LOGIN_04 | SQL injection attempt | Enter ' OR 1=1-- as username | Login is blocked TC_LOGIN_05 | Case sensitivity test | Enter uppercase username and correct password | Login behavior validated

Benefits for QA Engineers

This approach can help QA engineers by:

• Reducing repetitive test case writing
• Quickly generating test scenarios for new pages
• Assisting junior testers in identifying edge cases
• Speeding up the test planning phase

It does not replace manual testing but acts as a productivity assistant.

Limitations

Since the tool uses a lightweight local model, it has some limitations:

• Sometimes formatting needs minor adjustments
• It may generate generic scenarios
• Complex workflows still require human QA expertise

However, it provides a good starting point for test case creation.

Future Improvements

Possible enhancements for this tool include:

• Generating automation scripts automatically
• Exporting test cases to Excel
• Detecting different form types (login, signup, checkout)
• Generating security and accessibility test scenarios

Conclusion

AI is gradually becoming a powerful assistant for software testing.

By combining simple DOM analysis with AI generation, we can build tools that support QA engineers and reduce repetitive tasks.

This small experiment demonstrates how AI can assist in generating structured manual test cases directly from a webpage.

If you are a QA engineer exploring AI in testing, experimenting with small tools like this can be a great way to learn and innovate.

This is just the beginning! As a prototype, many more exciting features will be added in the near future.

Happy Testing!