Skip to content

Auto-Recording

Auto-Recording is the fastest way to build an automation in RTILA X. Instead of manually configuring every command, you simply browse the web normally — clicking, typing, scrolling, navigating — and RTILA X records every action you perform as a replayable automation command.

Think of it like a screen recorder, but instead of recording video, it records what you do so it can repeat those actions automatically, as many times as you want.

Browser with RTILA recording toolbar showing “Recording Actions…” indicator and Stop button


There are two ways to start a recording session:

Section titled “Method 1: From the Projects Page (Recommended)”

This method creates a brand-new project and immediately begins recording.

  1. Click Automation Projects in the sidebar.
  2. Click Record in the ribbon toolbar.
  3. A new project is created automatically with a spider-themed name (e.g., “Stealthy Jumping Spider - Data Extractor”).
  4. The Visual Builder browser opens with recording already active.
  5. The recording indicator bar appears at the top-right of the browser.
  6. Begin performing your actions — they are being recorded in real time.

If you already have an existing project open in the Visual Builder:

  1. Open your project in the Visual Builder (click Builder in the ribbon).
  2. Switch to the Web Automation tab.
  3. Click the Record button (video camera icon).
  4. The recording indicator appears and your actions begin being captured.

Tip: Method 1 is best for creating brand-new automations. Method 2 is best when you want to add more recorded actions to an existing project.


While recording is active, a red indicator bar appears at the top-right corner of the browser window. It contains:

  • A red pulsing background to remind you recording is active.
  • Text: “Recording Actions… (Alt+Shift+X to stop)”
  • A Start Loop button to begin a loop recording session.
  • A Stop button to end the recording.

The indicator pulses gently to keep you aware that everything you do is being captured.

Important: While the recording indicator is visible, every click, keystroke, scroll, and navigation you perform is being converted into an automation command. Be intentional with your actions.


RTILA X records a comprehensive set of browser interactions. Every action you perform is converted into a specific command type that can be replayed automatically.

Your Action Recorded Command Details
Clicking an element click Records the selector and position. Supports left, right, and middle mouse buttons.
Double-clicking click Recorded with click_count: 2 for double-click behavior.
Right-clicking click Recorded with button: "right". Useful for context menus.
Filling an input field fill Records the selector and the text you typed. Uses clear_before_fill: true by default.
Selecting a dropdown option select_option Records the dropdown selector and the chosen value.
Pressing special keys press Records keys like Enter, Tab, Escape, arrow keys, Backspace, Delete, Page Up/Down, Home, End. Also records keyboard shortcuts with modifiers (e.g., Ctrl+C).
Hovering over an element hover Only recorded if you pause on an element for 0.5 seconds or more. Quick mouse movements are ignored to prevent noise.
Scrolling scroll Records the scroll position after you stop scrolling (250ms debounce). Records both window scrolling and element-specific scrolling.
Dragging and dropping drag_to Records the source element, target element, and relative positions. Detected when you drag more than 10 pixels.
Selecting text select_text Records when you highlight/select text on the page.
Navigating via address bar goto Records the URL you typed in the address bar. Uses the browser’s webNavigation API to distinguish address bar navigation from link clicks.
Clicking a link wait_for_load_state Records a wait for the page to finish loading after a link click. Uses webNavigation transition type detection.
Switching browser tabs switch_to_tab Records which tab you switched to, including a fallback URL for reliability.
Resizing the browser window set_viewport_size Records the new window dimensions (width × height). Debounced at 500ms to avoid recording every pixel of a resize.
Action Reason
Interactions with RTILA’s own UI Clicking the recording indicator, extension panels, or refinement tools are ignored.
Hovering for less than 0.5 seconds Prevents noisy hover recordings from casual mouse movement.
Rapid scrolling (within 250ms) Debounced to capture only the final scroll position.
Interactions on html or body elements These are too broad to be useful as automation targets.
Typing regular text into inputs Regular typing is captured by the fill command when the field loses focus, not character-by-character.

There are three ways to stop a recording session:

Operating System Shortcut
Windows / Linux Alt + Shift + X
macOS Option + Shift + X

This is the fastest way to stop. The shortcut works regardless of which element on the page has focus.

Click the red Stop button on the recording indicator bar at the top-right of the browser.

  1. Open the RTILA side panel (click the RTILA extension icon in the browser toolbar).
  2. Switch to the Web Automation tab.
  3. Click the Stop button (which replaces the Record button while recording).

When you stop a recording, a summary popup appears on the page:

“Recording Stopped” You have recorded X command(s). Please click the RTILA extension icon in the top right corner of your browser to review and save your project.

This popup tells you exactly how many commands were captured and guides you to the next step.

Tip: After stopping, open the side panel to review your recorded commands. You can edit, reorder, or delete any command before saving.


Tip Why It Matters
Click precisely on elements Clicking the exact element produces more reliable selectors than clicking near an element.
Wait for pages to load Give pages time to fully load before interacting. RTILA records wait_for_load_state for link clicks, but manual navigation needs you to wait.
Use the address bar for navigation Typing URLs in the address bar produces clean goto commands. This is more reliable than clicking links for direct navigation.
Pause before hovering Hover commands are only recorded if you pause on an element for 0.5+ seconds. Move slowly and deliberately when you want a hover recorded.
Avoid rapid clicking Give elements time to respond between clicks. Rapid clicking can produce unreliable selectors.
Scroll deliberately Scroll to the position you want and stop. The recorder captures your final position after a brief pause.
Don’t interact with RTILA’s UI Clicks on the recording indicator, extension panels, or refinement tools are ignored and won’t be recorded.
Resize before recording If you need a specific window size, resize the window before starting your actions. The recorder captures the final size.
Test in a clean tab Start recording in a fresh tab to avoid interference from existing page state.
Use descriptive pauses When performing a sequence of actions, add brief pauses between them. This helps the recorder distinguish separate actions.

While recording, you can trigger RTILA’s AI to automatically extract data from the current page. This is perfect for when you’re recording a navigation sequence and want to add data extraction without manually building datasets.

  1. While recording is active (or even while not recording), right-click anywhere on the page.
  2. In the context menu, select “Auto-Extract Data with AI”.
  3. RTILA will:
    • Scroll through the entire page to load all content.
    • Annotate all visible elements with numbered identifiers.
    • Take a full-page screenshot.
    • Send the annotated screenshot to the AI model.
    • Generate datasets and automation commands automatically.
    • Add the generated datasets and commands to your project.

The AI analyzes the page screenshot and identifies:

  • Data structures — Product cards, tables, lists, article grids.
  • Repeating patterns — Elements that repeat (product cards, search results).
  • Interactive elements — Buttons, links, forms, navigation.
  • Pagination patterns — Numbered pagination, “Load More” buttons, infinite scroll indicators.
  • Generated datasets appear in the Data Extraction tab.
  • Generated commands appear in the Web Automation tab.
  • You can edit, modify, or delete any AI-generated dataset or command.
  • A confirmation popup tells you when the AI has finished processing.

Tip: AI Extraction works best on pages with clear, repeating data structures (product listings, search results, tables). For highly dynamic or heavily JavaScript-dependent pages, you may need to manually adjust the generated selectors.

If the AI detects pagination patterns on the page, it will ask:

“Pagination Detected” We detected potential navigation or pagination actions on this page. Would you like RTILA to automatically generate and append the automation logic (commands) for it?

  • Click “Yes, Add Automation” to generate commands for navigating through multiple pages.
  • Click “No, Just Data” to extract data from the current page only.

Understanding how recording works can help you get better results:

When you click an element, RTILA generates a robust selector using a priority system:

  1. data- attributes* — Most stable. Example: [data-product-id="123"]
  2. aria-label — Accessible labels. Example: [aria-label="Add to Cart"]
  3. role — ARIA roles. Example: [role="button"]
  4. name — Form element names. Example: [name="email"]
  5. ID — If the ID is not dynamic. Example: #main-content
  6. CSS classes — Safe, non-dynamic classes. Example: .product-card
  7. XPath fallback — For complex nested structures.
  8. Text matching — For elements identified by their text content.

The generated selector is formatted as css=selector,xpath=expression,text=content to provide multiple fallback options.

RTILA automatically detects and avoids dynamic selectors that are likely to break:

  • Pure numbers (e.g., 12345)
  • Random alphanumeric strings (e.g., aB3kF9x2)
  • UUIDs (e.g., f47ac10b-58cc-4372)
  • Very short class names (2 characters or less)
  • Classes with no vowels (likely auto-generated)
  • Classes starting with css-, sc-, tw-, r-, or Mui

RTILA can record interactions with elements inside Shadow DOM components. If you click an element inside a Shadow DOM, the recorder pierces through the shadow boundary to generate a working selector.

If you interact with elements inside an iframe, the recorder captures the iframe selector separately. The recorded command includes an iframe parameter so the automation can target the correct frame during playback.


Once you’ve stopped recording and reviewed your commands:

  1. Review commands — Open the side panel and check the Web Automation tab. Verify all commands were captured correctly.
  2. Edit commands — Click any command to edit its parameters, selectors, or display name.
  3. Add data extraction — Switch to the Data Extraction tab to define what data to scrape.
  4. Test the automation — Click Run to test your automation.
  5. Save — Click the Save button in the side panel to persist your project.

Problem Solution
Recording indicator doesn’t appear Wait a moment after clicking Record. The indicator may take a second to appear.
Actions aren’t being recorded Make sure you’re interacting with the page content, not RTILA’s UI elements.
Clicks produce wrong selectors Try clicking more precisely on the target element. Avoid clicking near element boundaries.
Hover isn’t recorded Pause on the element for at least 0.5 seconds. Quick mouse passes are ignored.
Scroll position isn’t captured Scroll to your desired position and wait briefly. The recorder debounces scroll events.
Right-click recorded but I didn’t want it Right-clicks are recorded as click commands with button: "right". Delete the command if not needed.
Tab switch not recorded Make sure you’re switching between actual browser tabs, not just clicking different areas of the page.