Risk-sized PR runs, rerun history, and run-finished notifications

Highlight
Featured this release
PR reviews size their test runs to the risk of the change
Instead of a flat 7–12 tests on every pull request, the review agent scales coverage to how risky the change actually is.
The PR review agent now scores each pull request's test risk from the change's blast radius, how central the affected flows are and what coverage already exists.

Low-risk changes get 1–3 tests instead of 7–12, while high-risk ones get the full 8–12. There is always at least one real test on the primary path, and the review summary carries a one-line note on the risk read and what it covered. It reads as coverage, never as a code-quality judgment, and the pass/fail verdict is unchanged.
# Direct link
Update 02 of 04
Get a browser notification when a run finishes
A bell next to the Running count arms a desktop notification and a chime for the run or test you're watching.
Click the bell – on the run results page, a single test result page or the test editor header – and QA.tech asks for browser notification permission once, then remembers it, so later runs are armed without clicking again.

The notification fires only when the tab is hidden or the window is unfocused, and its text carries the outcome, for example "17 passed, 1 failed". Clicking it focuses the tab back on the run. For Slack and email, see notifications.
# Direct link
Update 03 of 04
Move test cases back to Draft from chat
The chat agent could already activate and archive tests; it can now convert Active tests back to Draft too.
Ask the chat agent to move a test to Draft and it applies the change in place.

If other Active tests depend on that one – a login or precondition test, directly or through a chain – the agent names those dependents so you can include them in the same request or change their dependencies first. Activating a Draft test works exactly as before.
# Direct link
Update 04 of 04
Rerun history on the run results list
Every test row carries a badge that opens the full attempt history, with a link to each individual execution.
Click the ↻ N badge next to a test in the Results List to see every attempt with its status, duration and whether it was an automatic retry, plus a link straight to that exact execution.

Opening an older attempt shows a "Viewing Original run" notice with a "View latest rerun" link back. This brings the results page in line with the rerun history added to the chat panel earlier in August. See tests and results.
# Direct link

PR reviews size their test runs to the risk of the change

Get a browser notification when a run finishes

Move test cases back to Draft from chat

Rerun history on the run results list

Your team moves fast. Can your testing keep up?