v2026.08.03
TL;DR
Thunders AI keeps a browsable chat history and can bulk-create test cases, you can watch a test run while it executes, secret variables are encrypted at rest, and this image includes third-party dependency security patches.
FEATURE
- Thunders AI chat history: every conversation is kept, browsable, renamable, and resumable, instead of losing context between sessions.
- Bulk test case creation: ask Thunders AI to create several test cases in one request and review them all at once.
- Watch a run as it happens: open a test run while it is still executing and follow it live, instead of waiting for it to finish.
- "Close tab" step: tests can close a browser tab as an authored step.
IMPROVEMENT
- Native tool connections: Jira, Confluence, Azure DevOps, ClickUp, Shortcut, Linear, and Notion now connect directly to Thunders. Reconnect an integration, or add a new one, to switch to the new flow; every connected tool then becomes available to Thunders AI.
- New-version notice: the app now tells you when a newer version is running instead of erroring quietly in stale tabs.
FIX
- Secret variables: secret test variables are encrypted at rest and never sent back to the browser, and you can now change one after creation.
- Sequential execution: sequential runs now execute one test at a time instead of firing every CSV row at once.
- Xray Data Center: creating a test now tells you if some steps could not be added instead of reporting success, and Gherkin imports bring their scenario text across.
- Accessibility under strict security rules: accessibility checks work again on sites with strict content-security policies.
- Run reports: durations no longer count page load time twice, each step shows its own screenshot, and a failed check shows the screenshot it actually looked at.
- Steadier test runs: a batch of reliability fixes to waits, clicks, assertions, and URL matching that cut down on flaky failures.
- Dependency security patches: patched several third-party dependency vulnerabilities.
Ready to ship faster with smarter testing?
