Skip to main content

User Actions

The Autonomous Natural Language Processing (NLP) engine allows you to effortlessly create a custom flow test using simple, everyday language. This powerful technology lets you define steps using common English phrases like "Browse to http://example.com/" or "Click the OK button" without any coding required. You can include variables or parameters in a test to allow you to run the test in different scenarios.

As you enter instructions for a step, Autonomous provides helpful suggestions to guide you through the process.

Auto-Correction​

The system automatically corrects syntax errors and divides complex steps into multiple simple steps, making them easier to read and understand. The Auto-Correct feature makes the following corrections to commands:

  • Grammar and Spelling – For example, "Clack OK" will automatically be corrected to "Click OK".
  • Break Down Complex Steps – Steps with multiple actions will be broken down into shorter, more manageable steps. For example, "Enter 'John' in 'User name' and click OK" will be divided into:
  1. Enter 'John' in 'User name'
  2. Click 'OK'
  • Simplify and Standardize Commands – Steps will be modified to ensure consistency and clarity across all steps. For example, "In the user name field, enter 'John Smith'" will be changed to "Type 'John Smith' in the 'User Name'".

If steps were corrected, the icon indicates the number of corrections made. Hover over the icon to view details of the changes.

Examples​

Below are examples illustrating how easy it is to create tests using Natural Language Processing. Notice how the commands are written in straightforward English, making them accessible even to those without programming or technical expertise.

Fill Out a Form​

Go to https://sandbox.applitools.com/
Click "Form"
Type "Jane Smith" in the "Your Name" field
Enter "janesmith@example.com" in the "Email" field
select "Technical Assistance" from dropdown below "Contact Reason"
Press ENTER
Click "Submit"

Open E-Commerce Page and Add Item to Cart​

Go to https://sandbox.applitools.com/
Click "E-Commerce App"
Click "Outdoors"
Click "Small Succulent Planter Pot"
Select the second option from the "Variants" dropdown
Click "Increment"
Click "Add to cart"

Validate a PDF Document​

Open a PDF directly from a URL, navigate page by page, and run visual and textual checks — all in plain English. This is useful for validating business-critical documents like invoices, contracts, reports, and confirmation letters.

Go to "https://your-app.com/documents/sample-invoice.pdf"
Verify that the first page contains "Invoice #2024-001"
Verify that the first page contains "Total Amount"
Visually check this screen
Go to the second PDF page
Verify that the second page contains "Payment Terms"
Verify that "Net 30" appears on the second page
Visually check this screen
Go to the last PDF page
Verify that the last page contains "Authorized Signature"
Visually check this screen

Custom Flow editor showing a PDF test (steps panel on the left, PDF rendered in the interactive browser on the right)

For more on what PDF testing can do, see PDF Testing.

Entering Instructions​

When entering an action, identify a screen element, such as a button, text box, or image, and then specify what to do with that element, such as click, hover, or enter text.

Elements​

Elements on the page can be identified in various ways, including:

  • Text label, for example, "Get Started" button.
  • Position in a list, for example, "The third item in the Language list".
  • CSS Selector or X-Path selector ID, for example, "gh-minicart-hover".
  • Position in relation to another item, for example, Checkbox to the right of "I agree".
  • Location on the page, for example button at the top of the page.

Screen elements with more than one word should be in quotation marks ("...").

Common Commands​

Here are examples of common commands that the NLP engine supports. This is not a complete list, as there is no limit to the types of commands you can create.

PDF Testing​

Custom flow tests can validate PDF documents alongside web pages. Use this to verify short, business-critical documents such as invoices, contracts, statements, reports, and confirmation letters.

How It Works​

With PDF testing, you navigate to a PDF the way your users do and test it just like a web page — with Visual AI, text assertions, and interactions. When you open a PDF, Autonomous displays it in the interactive browser. The document renders consistently regardless of the browser or device selected for the test, so results aren't affected by viewer-specific differences.

There are two ways to open a PDF in a test:

  • As the Start URL – enter a direct PDF URL (for example, https://your-app.com/documents/invoice.pdf) when you create the test.
  • By navigating to a PDF mid-flow – a Go to step, or a Click on a link that resolves to a PDF, is routed through the PDF viewer.

Good to Know​

  • For a PDF, Full screen and Viewport both capture the current page.
  • Autonomous validates the document's content, not how different browsers' PDF viewers display it. PDFs render in one dedicated environment (stable run-to-run), so there's no browser/device matrix to configure.
  • Scanned PDFs are not supported.

A PDF rendered in the interactive browser, with page-navigation controls visible

What's Supported​

  • Text assertions on specific pages, for example Verify that the first page contains "Total Amount". See PDF Pages in Assertions for more examples.

  • Page navigation by ordinal or number, for example Go to the second PDF page or Jump to the 5th page. See the PDF page navigation command in Common Commands.

  • Visual checkpoints on the current page using Visually check this screen.

  • Form interaction — for PDFs with fillable form fields, you can type into fields and click controls using the same NLP commands as web forms (for example, Type "John Smith" in the "Full Name" field, Click the "Submit" button).

  • Zoom, fit-to-width, fit-to-page, and Ctrl/Cmd+F search in the interactive browser during authoring. These actions are not recorded as test steps.

Masking Secret Test Data​

If a step includes sensitive information, such as a password, PIN number, credit card number, or a URL with basic authentication, when you save the test, the sensitive data is stored encrypted on the server and masked in the interface. If the field is not automatically masked based on the name of the field, enter "Secret" before the field value to indicate that it should be masked. Once a secret value has been masked, there is no way to retrieve the value from the system.

In the following examples, the bold text will be masked:

Type "My-PASS" in the password field

Type secret "My-PASS" in "Value1" field

Navigate to "https://Username:**My-PASS**@autonomous-demo.applitools.com/basicAuth"

Upload File​

Autonomous supports file upload steps in custom flow tests. Upload your files to the File Management Dashboard to manage them, then start authoring tests that include file upload steps. When you create a file upload step, the system automatically detects that you need to upload a file and prompts you to select a file from the dashboard or upload a new one. Once you've uploaded a file, you can reuse it across tests without needing to upload it again.

You can:

  • Select files you have already uploaded during test authoring
  • Reuse these files across tests

Add a file upload step during test authoring​

  1. Create or open a custom flow test (go to Tests, click Create a new test, then select Custom flow; or click an existing test name).
  2. In the test editor, enter a Start URL and a Test name, then type or record the steps that lead up to the file upload action.
  3. When the step requires a file upload, click your application's upload button or file input in the interactive browser. Autonomous intercepts the native picker and shows the File upload detected modal.

File upload detected modal

  1. In the modal, do one of the following:
    • Select an already-uploaded file from the list.
    • Click Manage your files to open the File Management Dashboard in a new tab, upload your file there, then return to the test tab and select the file.
  2. Confirm your selection to upload the file to your application.
  3. Continue authoring the remaining steps, then click Create & run (new test) or Update & run (existing test).

NLP command examples​

You can also type file upload steps directly into the step text box using natural language. Supported phrasings include:

upload file "contract_signed.pdf"
attach file "employee_data.csv"
attach file "employee_data.csv" from the file input
select file "project_plan.docx"
select file "project_plan.docx" in the file chooser
add file "readme.md"
import file "dataset.csv"
pick file "server_logs.log"
pick file "server_logs.log" in the file picker
choose file "resume_final.pdf"
choose file "resume_final.pdf" in the file picker

Native Validation Support​

Autonomous preserves any constraints declared on the file input element. For example, if the input specifies accept="image/png,application/pdf", only files of those types are accepted.

warning

You can upload only one file at a time.

XPath Elements​

When recording a test in the interactive browser, there may be rare situations where the recorder is unable to assign a clear English name to a screen element. In this case, the recorder will identify the element with its XPath, which may be a long string of code, such as /HTML[1]/BODY[1]/DIV[5]/DIV[1]/MAIN[1]/DIV[2]/DIV[3]/DIV[4]/FIELDSET[1]/DIV[1]/DIV[3]/LABEL[1].

This can make the test more difficult to read; however, the test will still run successfully, and regular functionality, such as self-healing, will be applied to the element.

If you need to refer to the element in other steps in the test, you should identify it with the XPath. For example:

Click the "OK button" with the XPath selector "/HTML[1]/BODY[1]/DIV[5]/DIV[1]/MAIN[1]/DIV[2]/DIV[3]/DIV[4]/FIELDSET[1]/DIV[1]/DIV[3]/LABEL[1]"

Advanced Options​

A custom flow test can also include the following advanced commands: