Changelog 2026.05

Browser Script Locators

Browser script steps now use chained Playwright-style locators to pinpoint elements on a page. Instead of relying on a single CSS or XPath selector, you can compose multiple locators together, with each one refining the match from the previous, and end the chain with a First, Last, or Nth modifier when more than one element might still match.

Since Browser Bots use Playwright behind the scenes, Playwright’s full set of locator types is now available:

• Selector – CSS or XPath, the same flexible selectors you’re used to.
• Label, Text, Alt Text, Title, Placeholder – match by visible or semantic text.
• Test ID – match by data-testid, usually the most stable option.
• Role – match by ARIA role and accessible name.
• iFrame – step into an <iframe> and continue locating elements inside it.
• Filter – narrow an earlier match by text content or visibility.
• First, Last, Nth – pick a specific element when multiple still match.

Using locators makes browser scripts more resilient to small changes in markup, and brings them in line with how Playwright scripts already locate elements.

Your existing scripts using selectors are still supported as before.

Browser Recorder Improvements

The Loadster Recorder browser extension has been overhauled to take advantage of the new locator chains. While recording, it now generates Playwright-style locator chains (including iFrame entry) rather than a single CSS selector per step, so the scripts you capture are easier to read and more reliable to replay.

The autocomplete in the locator editor also got some usability improvements: you can now navigate it with the arrow keys, Tab, and Enter, and chains can be collapsed when they get long.

Delayed Registration & Social Login

The login and registration flow has been streamlined so you can start scripting in Loadster without creating an account first. You only need to register when you actually run a load test or start a monitor — until then you can play with the script editor, record scripts, and try things out.

The registration page itself has been redesigned and now supports social login with Google, GitHub, and Microsoft, on top of the existing email/password signup. There’s no need to remember another password if you already use one of those providers.

If you have an existing Loadster account with password and would like to switch to social login, it should work seamlessly as long as the email address from your social account matches your Loadster account. If you try this and experience any weirdness please let us know so we can merge your accounts.

Duplicate Bot Group

The scenario editor now has a duplicate button next to each bot group (population), so you can clone a group with all of its settings — bot type, count, region, ramp-up, script, datasets, and so on — and tweak the copy instead of rebuilding it from scratch. Handy when you want to run the same script from several regions at once or compare two slightly different load profiles.

New Wildcard Cloud Region: Asia (Any)

Load tests can now target a new Asia (Any) wildcard region, similar to the existing wildcard regions for other continents. Loadster will pick an available Asia region at launch time, which is useful when you don’t care exactly which Asia region runs the test as long as the latency profile is roughly right. Wildcard regions are especially useful for unusually large load tests that might exceed the capacity that AWS has on hand at any given moment, so you can let Loadster find cloud capacity for you instead of trial-and-error launches.

Monitor Pass Rate & Cycle Details

The monitor pass rate is now shown to three significant digits instead of being rounded, so a monitor that’s running 99.971% reads as 99.971% rather than 100%. Hovering the value reveals a tooltip with a breakdown of cycle statuses (pass, fail, error, and so on), making it easier to see what’s behind the number at a glance.

Graph Improvements

Stack charts in the dashboard now stack their series correctly, and all charts gained a horizontal helper line that follows your cursor to help with reading off values across multiple series at the same time point.

Other Improvements & Bug Fixes

• Capacity reservations are now released as soon as a cluster launch fails, rather than waiting for the normal timeout. This frees the capacity for the next attempt and shortens the reservation window.
• Fixed the team invitation form so that a browser autofill quirk can no longer cause repeated invitations to be submitted automatically.
• Fixed the onboarding helper widget, which had stopped allowing pointer events after dismissal and was making the page un-clickable.
• Fixed an OAuth signup path that was leaving customers with no Stripe customer record, which broke the billing page on first visit.
• Fixed a bug with social signup that prevented some new accounts from receiving their starter fuel.