Skip to main content
Propolis Builder offers two methods for test creation, each suited for different workflows.

​
Creation Methods

​
Visual Test Builder

Create tests manually with precise control over every action.Best for:
  • Complex workflows requiring specific logic
  • Tests with conditional steps
  • Precise assertion requirements
1

Create test suite and add test

Navigate to Test Suites → + New Suite or select an existing suite.Inside the suite, click + Add Test and provide:
  • Test name - Descriptive name (e.g., “User can complete checkout”)
  • Starting URL - Where the test begins
  • Browser Profile (optional) - Select if authentication is needed
2

Build test steps

Click + Add Step to add steps to your test. Choose from AI steps (for interactions), Check steps (for assertions), and Wait steps.See “Available Step Types” below for details on each type.
3

Run and debug

Click Run Test. If it fails, use the Agent Execution Breakdown to see exactly where it went wrong, then edit your steps and run again.
4

Save

Once passing, your test is automatically saved in the suite and ready to run on-demand or on a schedule.

​
Available Step Types

The most flexible and powerful step type. Describe what you want to accomplish, and the AI figures out how to do it.Configuration:
  • Goal: What should this step accomplish? (e.g., “Click the submit button”, “Fill in the email field with a test email”)
  • Expected Outcome: What should be true after the step completes? (e.g., “The form is submitted”, “The email field contains a value”)
Examples:
  • “Click the login button”
  • “Enter a valid email address into the email field”
  • “Select the first available option from the dropdown”
  • “Scroll down to the pricing section”
Use when: Most scenarios - clicking, typing, selecting, scrolling, any user interaction. The AI adapts to UI changes automatically.
Pause execution for a specific duration.Configuration:
  • Duration: Time to wait in milliseconds (e.g., 1000 = 1 second)
Use when:
  • Waiting for animations to complete
  • Giving time for background processes
  • Debugging timing issues
The AI automatically waits for elements and page loads, so you usually don’t need explicit wait steps.
Verify that something is true about the current page state.Configuration:
  • Check Prompt: Describe what should be verified (e.g., “A success message is displayed”, “The URL contains ‘/dashboard’”)
Examples:
  • “A success message is displayed”
  • “The page shows the user’s profile information”
  • “The cart total is greater than $0”
  • “The login button is no longer visible”
Use when: Validating test outcomes, confirming page state, asserting expectations.

​
Troubleshooting Manual Tests

Common causes:
  • Authentication required: Make sure you’ve selected a Browser Profile with login state
  • Goals too vague: Make AI step goals more specific (e.g., “Click the blue submit button” not “Click button”)
  • Wrong starting URL: Verify the starting URL in your test configuration is correct
  • Timing issues: Add wait steps between actions if elements load asynchronously
If the agent can’t locate elements:
  • Make your AI step goal more descriptive (include color, position, or surrounding context)
  • Check if the element is inside a modal or hidden section that needs to be opened first
  • Verify the page loaded completely before the step executes (add wait step if needed)
If check steps aren’t passing:
  • Make the check prompt less strict - focus on what actually matters
  • Ensure previous steps completed successfully before checking
  • Add a wait step before the check to give the page time to update