Thunders News Center
Match a text exactly in a step
You can now tell the runner that a text in a step must match exactly. Mark the text as exact, and the runner looks only for an element with that exact text, with the same words and the same case. It does not choose a similar label. When the text is not on the page, the step fails with Exact Text Not Found.
This helps when a page shows labels that are close to each other, such as Save and Save as draft, or when the case of a word matters.
- Mark a text in the step editor: in an AI step, select the words and choose Mark as exact text, or click Add exact text and type the text. The text appears as an Exact match chip in the editor and in the step list.
- Write it yourself: type exact:"Sign In" in an AI step or in a condition. A variable inside the quotes also works.
- Use it where a step finds an element: clicks, typing into a field, hover, drag and drop and the other element actions, wait steps and verify steps.
Plain double quotes keep a loose match. Add exact: when the text must match strictly. On mobile apps, the exact match ignores case. Thunders AI and the recorder do not add the exact marker for you, so add it to the steps that need it.
#Sprint36
Debug one data file row in a live run
You can now run one row of a data file live from the test case page, with the values of that row. Breakpoints, pause on failure, step edits and resume then work as in any live run. Before, the rows of a data file ran only as queued runs, which cannot pause.
- Pick the row: open the menu next to Execute, choose the row in the picker under Execute row N live, then click Execute row N live. When the file has more than 8 rows, search the picker.
- Debug with the values of the row: set a breakpoint, pause on a failure, fix a step and resume. The run keeps using the same row.
- See what the row controls: the settings that the row sets, such as the browser or the resolution, are locked in the settings bar while the row is in use. They are never saved as the settings of the test case.
- Run every row as before: Execute all rows starts one queued run for each row.
The row stays in use while its browser is open. End Session, or a plain Execute, brings back the settings of the test case. While a row run is running or paused, you cannot change the data file of the test case.
#Sprint36
Skip a step of a reused test case in one test case only
When you reuse a test case inside another one, some of its steps may not fit the new context. You can now skip a step of the reused test case for the current test case only. The original test case, and the other test cases that reuse it, still run that step.
- Skip one step: expand the reused test case, open the More actions menu of the step, and choose Skip in this test case. Choose Enable in this test case to run it again.
- Skip several steps at once: select steps of the reused test case and use the bulk action bar. This also works for a reused test case nested inside another one.
- Keep your choice in copies: when you copy and paste these steps, or create a test case from a selection, the skipped steps stay skipped.
A skipped step appears greyed out, like other skipped steps. When the whole reused test case is skipped, enable it first before you change one of its steps. To delete or edit a step for every test case, open the original test case.
#Sprint36
Organize and preview test assets
Test Assets now gives data sets and test files their own place, so the material your tests depend on is easier to find, check and reuse. You can organize both in folders, look inside an asset without downloading it, and see where it is used.
- Manage data sets and test files separately: each asset shows its type, so a data set with rows is not mixed up with a file such as a PDF or an image.
- Organize assets in shared folders: group related material together, so the whole project finds it in the same place.
- Preview the contents: open an asset to see its rows and columns before you use it in a test.
- See which test cases use an asset: the Used in tab lists them, so you know what a change to the asset can affect.
- Save files downloaded during a run: keep a file that a run downloaded in Test Assets, and reuse it in other tests.
You can also upload test assets up to 250 MB. This upload limit does not change the separate limits of features such as file comparison.
#Sprint35
Choose the browser size you need
The preset resolutions do not always match the screen your design targets. You can now set a custom browser resolution for a web test, so the test runs at the width and height you need.
- Open Custom resolution: in the browser settings of a test, click the button next to the resolution list.
- Pick a format or go free: choose 16:9, 16:10, 4:3 or 21:9 to keep the ratio, or choose Free to set the width and the height independently.
- Enter the width and the height in pixels: any size works within the 15-million-pixel limit. Apply it, and the test uses that resolution.
Matching the intended viewport helps you prepare a visual comparison. It does not make differently sized references equivalent, so capture the reference at the same resolution.
#Sprint35
Manage more of your tests from an MCP-connected AI client
An AI client connected to Thunders through MCP can now handle more of the setup around a test, rather than stopping after creating its steps. The additions cover test-case details, saved execution settings and the device and build information needed to queue a mobile run.
- Update an existing test case: change its name, description and status, or set its saved environment and persona. The same permissions apply as in the web app; the browser is chosen when queuing the run rather than through the saved-settings tool.
- Prepare a mobile run: list available device pools and models, inspect the build attached to an environment, and attach an existing build from another environment in the same project.
- Repeat a test case up to five times in one queue request: repetitions can start concurrently, including repeated copies of a sequence. Choose parallel execution with a maximum parallelism of one when they must run one at a time.
Build attachment reuses a build that is already in Thunders. Upload the first build through the web app or CI; these MCP additions do not upload a new build file. Mobile tools are available on deployments with mobile testing enabled.
#Sprint35
Accessibility reports with element images and remediation guidance
RGAA reports now provide more context for investigating a finding and deciding what to change. Element captures help locate the affected part of the page, while remediation guidance in the web RGAA PDF brings useful correction details into the report you share with others.
- Identify the affected element: reports include element images where a capture is available. When an element cannot be captured, an explanatory tile appears instead of implying that an image exists for every finding.
- Read the suggested correction: non-compliant rules with an entry in the correction catalogue include the impact, recommended fix, before-and-after code and reference links in the web RGAA PDF. This guidance is not added to every rule, to “To verify” findings, or to every report format.
- See what the scan could not assess: the RGAA audit grid lists images that could not be tested, so those gaps remain visible for follow-up rather than disappearing from the report.
This release also addresses specific accessibility findings in Thunders itself, including contrast, accessible control names and page structure on the Test Case page, the Thunders AI page and the shared layout. These are targeted corrections, not a claim of complete RGAA compliance.
#Sprint35
Work with groups of test steps more easily
The test editor now makes it easier to focus on the part of a test you are changing and apply the same action to several steps. Reused cases can be collapsed when you only need the outline, while selection and keyboard shortcuts reduce repeated menu actions.
- Expand or collapse a reused test case: show its nested steps when you need the detail, or hide them to follow the parent flow. The controls are available in both the editor and the run viewer, including for nested reused cases.
- Skip or enable several steps together: select the relevant steps and use the bulk action bar instead of changing them one at a time.
- Copy selected steps with Cmd/Ctrl+C: the shortcut acts on checked steps when you are not editing a field or selecting ordinary text. Copied steps can be pasted into the same test case or another case.
- Review before deleting: Backspace or Delete opens the confirmation for the selected steps; it does not immediately remove them. This deletion shortcut does not apply to Discovery steps.
#Sprint35
Understand access problems before you start testing
An app that cannot be reached is not always asking for a password. Access checks now distinguish authentication failures from forbidden access and detected bot or WAF challenges, giving you a more relevant next step before you start testing.
- Try Stealth Mode for a detected bot challenge: the access-check card offers a retry with Stealth Mode. It can help with detection-related blocks, but it does not guarantee access or replace the app’s firewall and authentication requirements.
- Tell 401 and 403 responses apart: a 401 points to authentication; a 403 gets separate guidance for access restrictions rather than asking for credentials by default.
- See certificate warnings while recording: Chromium recordings can warn when a page has an invalid certificate, and certificate errors are reported for the affected pages in recordings and runs.
Deleting an environment from its settings page now also works correctly, without requiring you to return to the environments list.
#Sprint35
Identify the data row behind each test result
When the same test runs against several rows of a data file, its name alone is not enough to explain the results. Each run now shows which data row it used, so you can distinguish a failure for one record from a failure across the whole test.
- Find the row in the runs list and run panel: row information stays alongside the execution result, rather than leaving you with several identically named runs.
- Keep that context in the PDF report: a shared report identifies the data row behind a result, making it easier to discuss a specific failure without reconstructing the run from the original file.
For example, if a test checks the same flow for several records, you can use the row information to identify the record that failed and open the corresponding run for investigation. The test remains reusable; its results no longer depend on the test name to tell every execution apart.
#Sprint35