Latest News from the Thunders Product Team

Thunders News Center

New feature

Set a higher monthly credit limit for busy testing periods

Ines avatar
Shared by Ines •

When testing demand exceeds the credits included in your plan, an Admin can now raise the organization's monthly credit limit from the Billing tab. You can allow additional usage without waiting for the next monthly period, while keeping a limit on how far that usage can go.

The setting is available to eligible active, paid Starter and Pro organizations on monthly or yearly subscriptions. Trial, past-due and group-managed subscriptions do not use this setting.

  • Open Billing and edit Monthly credit limit to choose an amount above your plan's included credits.
  • Set the limit as high as 10 times the plan's monthly credits, or return it to the plan allowance when you no longer need the extra headroom.
  • Continue testing within the configured limit. Usage above the credits included in the plan is billed as overage; raising the limit is not an allocation of free credits.
  • Follow consumption against the configured limit in the Usage tab.

#Sprint34

Improvement

More automatic RGAA checks and clearer accessibility reports

Ines avatar
Shared by Ines •

Accessibility auditing involves both checks that software can perform and judgments that still need an auditor. Thunders has expanded its automatic RGAA 4.1 checks and made the results easier to interpret, so you can review the findings with a clearer view of what was actually checked.

The work covers additional checks across areas such as images, links, page structure and forms. It does not turn an automated scan into a complete accessibility audit or a declaration of compliance.

  • Review more automatically evaluated RGAA tests in the accessibility report, while keeping human review for cases that require it.
  • See rule counts that reflect the report's actual results rather than misleading totals.
  • Read the exported RGAA audit grid and PDF without the AI confidence percentages previously included alongside results.
  • Avoid the specific screenshot timeout caused by trying to capture an element that is hidden at capture time: that screenshot is skipped instead of waiting for an invisible element.

#Sprint34

Improvement

Thunders MCP: continue a web debug run after changing a step

Ines avatar
Shared by Ines •

When you use Claude or another MCP client to investigate a failing test, a small step correction should not always require another execution from the beginning. Thunders MCP can now start a web debug run that keeps its browser open, then continue that run from the first changed or failed step.

This gives the client a way to check a correction in the existing browser state while the session is still available.

  • Start the test in debug mode through the MCP client, using its saved run settings.
  • After execution finishes, update the relevant step and ask the client to continue the run. Thunders identifies the continuation point rather than requiring a new queued run.
  • Validate a proposed step list before saving with the MCP's validation tool, and receive specific validation errors when steps do not meet the supported authoring rules.
  • Bind a data file to a test case through MCP for data-driven testing.

Continuation is for a single web debug run in Chrome or Chromium, not mobile or SAP execution. It requires a compatible browser session that is still open; inactivity or the session limit can close it, in which case a new run is needed.

#Sprint34

New feature

Run statuses that distinguish missing setup from a test verdict

Ines avatar
Shared by Ines •

A test that cannot start because its configuration is incomplete needs a different response from a test that finds a problem in your application. Run statuses now make that distinction clearer, and conditional steps can end a test with an outcome that matches the situation.

  • Not Runnable identifies missing prerequisites, such as an unresolved application URL, variable or file variable. The pre-run check lists the missing values so you can correct the selected environment before executing.
  • When authoring a conditional step with a Stop action, choose Success, Failed, Inconclusive or Skipped. For example, use Failed when an error condition is a product failure, or Inconclusive when an external dependency prevents a verdict.
  • The selected outcome applies when the condition is met. A false condition continues the test, and cleanup steps still run after a conditional ending.
  • Stopped is reserved for a person stopping a run. Existing conditional Stop steps without a selected outcome end new runs as Inconclusive; their past results are unchanged.
  • iOS run checks also identify unsupported Android-only actions, including Wi-Fi, mobile data and location-services toggles, and show the affected steps.

#Sprint34

New feature

Create API tests with your OpenAPI specification as context

Ines avatar
Shared by Ines •

Describing an endpoint from scratch is unnecessary work when its requests and responses are already documented. You can now attach an OpenAPI specification to an API app and use it as context when asking Thunders AI to create or edit API tests.

The setup keeps the specification with the app, while the conversation shows when Thunders AI reads the specification or makes a live API call. You can distinguish what comes from the documentation from what was observed in a response.

  • Upload a JSON or YAML specification, or provide its URL, from the API app setup. Thunders parses and stores it for later use.
  • Set the app's base URL and check its reachability during setup. When available, the specification's server information can help fill the URL.
  • Describe the scenario you want to test. Thunders AI can use the specification to prepare API calls and explicit assertions, and use a live probe to clarify a response when needed.
  • Review specification reads and live calls together in the conversation card.

This is an API-app workflow; attaching a specification is not a claim that every endpoint or business scenario is automatically covered.

#Sprint34

Feature update
Improvement

Authenticated Profiles: share a pool without sharing the same profile

Ines avatar
Shared by Ines •

Parallel web tests can interfere with one another when they reuse an authenticated session. Profile pools let you group interchangeable saved profiles and have Thunders allocate an available member to a run, instead of choosing a profile manually for each execution.

Profiles also have clearer scope. Keep a profile at project level for reuse, or define an environment-specific version when staging and production need different sessions.

  • Create a named pool and reference it from an inject-profile step. Thunders selects a free member and manages its reservation while it is in use.
  • When all members are busy, a run waits for availability rather than taking a profile already reserved by another run. Waiting is limited to 15 minutes.
  • Pool selection skips members whose saved session has been rejected by the application.
  • Manage profiles and pools in Settings, with visibility into project inheritance and environment overrides. An environment entry takes precedence over a project entry with the same name.

Authenticated Profiles are for web tests. To avoid account-level conflicts, populate the pool with suitable separate application accounts: separate saved profiles alone do not create independent accounts.

#Sprint34

New feature

Skills: reusable instructions for the way your team tests

Ines avatar
Shared by Ines •

Your team may have a particular way of turning acceptance criteria into tests, investigating a failed run, or describing a test step. Skills let you save those instructions in Thunders instead of adding the same explanation to each new conversation.

A skill gives Thunders AI guidance for the request you apply it to. Project members can reuse the same instructions, and you can refine them as your practices change.

  • Create a project skill with a name and natural-language instructions, then select it from the Thunders AI input when you need it.
  • Start with built-in skills for test ideas, tests from acceptance criteria, and investigating a failed run. Customize a built-in skill when your team needs a different approach.
  • View and edit the saved instructions from the skills screens, rather than maintaining copies in separate conversations.
  • Use the same library from an MCP client: connected agents can list and read skills, use their instructions, and create, update or delete project skills with the appropriate permissions.

Skills guide the work; they are not a guarantee that generated tests need no review.

#Sprint34

New feature

Review test changes on a branch before merging them

Ines avatar
Shared by Ines •

Preparing tests for a new feature should not mean changing the version your team relies on today. With branches, you can work on a separate version of your test cases and test sets, run it against the right environment, and bring those changes back to the main branch when you are ready.

The review and merge flow lets you compare the two versions before deciding what to keep. If work on the main branch overlaps with yours, you can resolve the conflicts rather than overwrite the other changes without seeing them.

  • Create a branch from the branch selector and switch straight into it to start editing.
  • Review the differences before merging, including which version comes from the main branch and which comes from your working branch.
  • Resolve conflicts by choosing the version to keep, and bring in changes from the main branch when yours has fallen behind.
  • Keep branch context with your test runs, so you can tell which version of the tests produced a result.
  • Target a branch through the Public API and Thunders MCP as well as in the interface.

#Sprint34

Mobile
Improvement

Find the right mobile device, and keep running

Ines avatar
Shared by Ines •

You pick a device once, and your runs should keep going. Choosing it is quicker, and a busy or missing device no longer stops you without a word.

  • Find the right device faster: sort the list, filter it by EU or US region, and see the brand in each device name.
  • Keep running when your pinned device is busy: Thunders takes a free device with the same name.
  • See a clear message when a pinned device is no longer available, instead of an empty run.
  • Start from a clear picker: with no device pool in your project, it shows you how to create one.

#Sprint33

Mobile
Improvement

Upload your mobile builds from your CI

Ines avatar
Shared by Ines •

If your app changes every day, uploading each build by hand does not scale. Your pipeline can now hand the new build to Thunders itself.

  • Upload an APK or an IPA from your CI with an API key, on GitHub Actions, GitLab CI, Azure DevOps or any pipeline that can send a file.
  • Replace the build of an environment, and change nothing in your test cases. Your tests point to the app, so the next run uses the new build.
  • Spot the builds that your pipeline uploaded: the build field shows a CI badge.
  • Know which build each run tested: the build version shows in the build field and in the test run report.

#Sprint33