Skip to main content

​
Overview

The AI Action Agent lives inside the test editor and automates your scenario notes into executable test steps. It reads your scenario notes and performs the described interactions in the live browser, creating steps as it goes.

​
Where AI Action Agent Lives

AI Action Agent is accessed inside the test editor, not from the global chat. The global chat is for the Test Planner Agent, which creates test plans and scenario notes.
AI Action Agent
Global Chat → Test Planner Agent (creates plans)
Test Editor → AI Action Agent (automates steps)

​
How It Works

  1. Open a test case in the editor
  2. Ensure you have scenario notes describing the test flow
  3. Start the live browser session
  4. Click the AI Action Agent button
  5. Watch as the agent performs actions and creates steps
AI Action Agent in action

​
The Automation Flow

Scenario Notes (what to do)
    ↓
AI Action Agent reads and understands
    ↓
Agent performs actions in live browser
    ↓
Steps are created automatically
    ↓
You add assertions manually

​
What AI Action Agent Can Do

The Action Agent excels at primary user interactions:
Action TypeExamples
ClickButtons, links, checkboxes, radio buttons, tabs
FillText inputs, textareas, search fields
SelectDropdown options, list selections
ScrollPage scrolling, scrolling to elements
NavigateFollowing links, page transitions

​
Example: Login Automation

Scenario Notes: 
1. Navigate to the login page
2. Enter the email address [email protected]
3. Enter the password securepass123
4. Click the login button
AI Action Agent creates:
  • Navigate step: /login
  • Fill step: email input with [email protected]
  • Fill step: password input with securepass123
  • Click step: login button

​
What AI Action Agent Cannot Do

AI Action Agent focuses on performing actions. It cannot add assertions or verification steps. You must add these manually.
Must be added manually:
  • Verify Text steps (checking text content)
  • Verify Visibility steps (checking element visibility)
  • Verify URL steps (checking page URL)
  • Visual Assertions (screenshot comparisons)
  • Any form of verification or assertion

​
Why Assertions Are Manual

Assertions require human judgment about:
  • What exactly should be verified
  • What the expected values should be
  • Which elements are important to check
The Action Agent automates the “doing” part of tests. The “checking” part remains your responsibility.

​
Using AI Action Agent

​
Step-by-Step Guide

  1. Prepare your test case
    • Create a test case with a descriptive title
    • Add scenario notes describing the flow
  2. Open the editor
    • Click on your test case to open it
    • Start the live browser session
  3. Run the Action Agent
    • Click the AI Action Agent button in the toolbar
    • The agent reads your scenario notes
    • Watch the live browser as actions are performed
  4. Review created steps
    • Steps appear in the editor as they’re created
    • Verify locators are correct
    • Check that actions match your intentions
  5. Add assertions
    • Use the recorder or element picker
    • Add Verify Text, Verify Visibility steps
    • Configure expected values

​
Example Workflow

Test Case: “User can add item to cart” Scenario Notes: 
1. Navigate to the products page
2. Click on the first product
3. Select quantity 2
4. Click Add to Cart button
5. Open the cart
After AI Action Agent runs: 
Step 1: Navigate → /products
Step 2: Click → product card
Step 3: Fill → quantity input with "2"
Step 4: Click → "Add to Cart" button
Step 5: Click → cart icon
You add manually: 
Step 6: Verify Text → cart shows "2 items"
Step 7: Verify Visibility → product name visible in cart

​
Writing Effective Scenario Notes

​
Be Specific

Less effective:
"Login to the app"

More effective:
"Navigate to /login, enter email [email protected],
enter password test123, click the Submit button"

​
Include Element Hints

Help the agent find the right elements: 
"Click the blue 'Add to Cart' button below the price"
"Fill the email field labeled 'Email Address'"
"Select 'Medium' from the size dropdown"

​
Describe the Flow Sequentially

1. Go to the homepage
2. Search for "laptop" in the search bar
3. Click on the first search result
4. Click the "Buy Now" button
5. Fill the shipping form fields

​
Troubleshooting

​
Agent Can’t Find Element

  • Check if the element is visible on page
  • Add more specific description in scenario notes
  • Verify the element isn’t inside an iframe
  • Try refreshing the live browser session

​
Wrong Element Selected

  • Update scenario notes with more specific text
  • Include the element’s label or nearby text
  • Specify position: “the first button” or “button at bottom”

​
Actions Not Matching Expectations

  • Review scenario notes for ambiguity
  • Break complex actions into smaller steps
  • Use explicit element descriptions

​
Agent Stops Mid-Flow

  • Check if a page navigation occurred
  • Verify no popups or modals appeared
  • Check for CAPTCHA or authentication prompts

​
AI Credit Usage

ActionCredits
AI Action Agent execution1 credit per action performed
Running the agent on a 5-step scenario uses approximately 5 credits.

​
Best Practices

​
Combine with Manual Steps

Use Action Agent for repetitive setup, then add manual assertions: 
AI Action Agent handles:
→ Navigate to page
→ Fill form fields
→ Click buttons
→ Select options

You add:
→ Verify success message
→ Check URL changed
→ Confirm data displayed correctly

​
Test Incrementally

Don’t automate everything at once:
  1. Automate 2-3 scenario notes
  2. Run and verify
  3. Add assertions
  4. Continue with next notes

​
Keep Notes Atomic

Each note should be one clear action: 
Good:
1. Click the login button
2. Enter email in the email field

Not ideal:
1. Click login and enter credentials