Version 1.72.0 (2026-09-28)

Changes breaking backward compatibility

• New Upgraded Chrome and Android WebView on arm64-v8a Android cloud devices — to v145 on Android 13–16, and to v133 on Android 8.1–12. WebView tests on older Android versions can now display modern web pages. Please note that this update may cause visual differences in your test results. (#472, #535)
• New Upgraded the Chrome version to 153, the Edge version to 153, and the Firefox version to 155 on cloud browsers. Please note that this update may cause visual differences in cloud browser test results.
• Fix Changed the policy that decides which URLs the "Web API call" command may request, for security. Please note that this may block existing "Web API call" steps that previously passed.
• Fix Fixed an issue where the "Terminate the app under test" step unnecessarily reset the iOS cloud device when the test was started with the "with device reset" launch option. Every such step spent 1–2 minutes clearing the device and could fail with "Failed to clear the device".

General

• New Shared steps can now be organized into folders. This feature is also available on the MagicPod MCP Server. (#43)
• New Failed test cases in a batch run can now be retried from the batch run detail page, through the MagicPod Web API, and through the MagicPod MCP server. (#406)
• Fix Reverted the wait limit for conditional steps such as "If UI element exists" and "If UI element is visible" from 20 seconds back to 10 seconds. The 20-second limit introduced in 1.71.0 made batch runs noticeably slower, because every such step whose UI element was not found waited twice as long. (An issue since 1.71.0)

Generative AI

• New MagicPod MCP Server now supports OAuth, which enables ChatGPT and some other AI products to use it.
  • Currently available only for email/password authentication.
• New Upgraded the model used by MagicPod Autopilot to Claude Sonnet 5.

Mobile App Testing

• New Added the "Pass biometric authentication" and "Fail biometric authentication" commands, which send a matching or a non-matching biometric scan to a biometric authentication dialog. They work only on cloud devices, and Android requires 14.0 or later. (#391)
• New The Google Play Store is now built into Android 16 and 17 arm64-v8a cloud devices, so apps that rely on libraries delivered through Google Play services, such as ML Kit, now work.
• New The "Set image for camera" command can now be used when testing Safari on iOS cloud devices.
• Fix Fixed an issue where the app under test could crash when it opened the camera during an iOS test. (An issue since 1.70.0)

Browser Testing

• New Playwright is now available for browser tests run on BrowserStack.
• New Updated the list of devices available for mobile emulation, adding recent models such as Pixel 10.
• Fix Fixed an issue where the "Navigate to URL" step failed with a "Download is starting" error in Playwright tests when the URL pointed to a file download.
• Fix Fixed an issue on iOS Safari where steps that rely on coordinates, including "Input keyboard key", could act on the wrong position once the software keyboard appeared.
• Fix When a step times out on the Playwright engine, the error message now states why the UI element could not be operated — for example, that it stayed disabled or invisible — instead of only reporting the timeout.
• Fix Fixed an issue where the "Select from dropdown menu" step could fail in Playwright tests when an option label had leading or trailing whitespace, or when the same label appeared more than once.
• Fix Fixed an issue where a WMV file uploaded in a cloud browser test was not recognized as a video file. (An issue since 1.70.0)