Release notes··5 updates
5 updates shipped on August 5, 2026.
Update 02 of 05
Stop freezes the response immediately and won't let a pending analysis or test run resume the conversation later.
The stop button used to wait on a network round trip before the interface caught up, while the current step's remaining tool calls kept running.
It now stops the moment you click it. If a site analysis, test run or crawl was in flight, it won't quietly resume the conversation afterwards – a short "complete" notice waits in the queue until your next message.
Update 03 of 05
Ask about a setting or integration a few clicks deep and the agent goes and finds it instead of reporting it doesn't exist.
Chat-initiated site analyses used to default to a shallow crawl that only reached pages linked from the homepage, so anything nested – settings, billing, admin screens – was out of reach and the agent could wrongly conclude a feature didn't exist.
Analyses that name a target now crawl deeper automatically, which means test discovery reaches those pages before the agent answers.
Update 04 of 05
A test created from a TestRail or Xray case now carries a badge back to it – groundwork for pushing results there.
When chat creates a test from an existing TestRail or Xray case, the new test case shows a "TestRail 4711" or "Xray PROJ-42" badge in your test lists, linking out to the original case where a URL is known.
Copying or duplicating a test leaves the link behind, so two tests never point at the same external case. Result reporting back to those tools isn't wired up yet – this release only tracks the connection.
Update 05 of 05
Testing an Xray connection shows the tests it found, not just a pass-or-fail toast.
An Xray Cloud API key is tied to a Jira account, so it can authenticate perfectly and still see none of your tests. Settings → Integrations → Xray now reads back a sample after "Test connection": instance URL, response time, and up to five tests with their Jira keys, summaries and folders.
That makes it obvious whether the connection reaches the right project before you start pulling cases into chat.
QA.tech agents test your product autonomously, so moving fast never means shipping broken. See how it works in a 30-minute demo.