v2026.09.03
TL;DR
Insert a breakpoint, execute, fix, and resume. Set your default test execution settings per environment, reference the running environment’s name from a step, upload Excel files for batch runs, and see why a run stopped even when no step failed.
FEATURE
- Breakpoints, pause and resume a run: set a breakpoint on a step and the run stops when it reaches it. Look at where the run is, edit a step behind the breakpoint, and resume from there instead of starting over.
- Run settings per environment: the environment now has web and mobile run settings tabs, so your preferred execution settings can be applied to all test when running with that environment instead of being set again on every test.
- A built-in ENVIRONMENT_NAME variable: every environment now comes with
[ENVIRONMENT_NAME]already defined. Use it in a step the way you use your own variables and it resolves to the name of the environment the run is using, so one test case can tell Staging from Production without you wiring anything up. It appears in the variables list as read-only, and the name is reserved so none of your own variables can clash with it. - Excel data files: batch execute accepts .xlsx alongside CSV.
- Restrict which execution clusters a project can use: choose the allowed clusters from project settings.
- Visual Compare, overlay or crops: switch between the full-page overlay and the per-difference crops.
- Guidance when a captcha blocks reaching your app: if the app check hits a captcha, Thunders now says that is what happened and can run a bypass script to get past it.
- Thunders opens your last project: the project you were last working in opens by default.
IMPROVEMENT
- A run that stops without a step failure now tells you why: a banner above the steps list gives the reason, instead of leaving you with a run that just stopped and no failed step to point at.
- Self-hosted parallelism is respected: a run honors the execution cluster's configured maximum parallelism on self-hosted deployments.
- Steadier Browserless connections: connection retries now back off when Browserless is rate limiting, with jitter and a shared time budget rather than hammering it.
- More complete PDF reports: unexecuted and skipped steps are reported, and durations show two decimals.
- More accessibility checks decided automatically: eleven RGAA tests are now decided mechanically instead of needing a manual verdict.
FIX
- A failed token renewal no longer signs you out: when a silent sign-in renewal fails, Thunders recovers it interactively instead of dropping your session.
- Steps added during a run stay: the runner no longer deletes steps that were added while the run was in progress.
- File fields keep one usable reference: File Compare and the Visual Compare image field each offer a single valid file reference instead of ambiguous options.
- Chrome's breach dialog no longer swallows input: a password-breach popup during a run stops eating the keystrokes meant for the page.
- Steadier test runs: conditional steps resolve their targets correctly, transient elements are re-triggered through a remembered hover, and vision coordinates scale proportionally on both axes.
- Thunders MCP: a custom MCP connector now explains why its sign-in failed, with a specific reason, instead of failing without telling you what to fix.
Ready to ship faster with smarter testing?
