A question register regenerated with 62 open, 44 overdue and 20 aged 70-73 days, and three suite-gating questions had been visible for eleven weeks. Visibility was never the blocker.
Three entries sat under "Answered Questions" while carrying "(working assumption)" and "(confirm in environment)". The open-questions collector only harvests what is filed as open — so they aged forever.
A textbook bug report — steps, screenshots, a live order link — was invalid because its Expected Result came from a field name rather than a specification. One extra template line catches that whole class.
A Custom Metadata field named Manager_Approve_Max_Pct__c stored an absolute £32 ceiling. The naming defect was rejected as cosmetic. Twenty days later it produced a Highest-priority bug against correct behaviour.
An invalid bug was cancelled and then unlinked from its story. The sprint's linked-bug count went from 102 to 101, and the only record that testing ever pushed back went with it.
Two Highest-priority defects raised nine minutes apart, both closed as invalid with the same sentence. The story shipped the product catalogue layer; the tester tested the screen and the document, which were correctly empty.
Four bug tickets moved in one day and no code was written. One travelled the full development path because a developer read a field name correctly; another was closed into a status named Ready For Sit Deployment with nothing to deploy.
One High-priority defect, 33 stories, 76 linked bugs — and two monitoring reports that describe worlds with nothing in common. An unlinked bug is not scored as low risk; it is absent.
A weekly rollup opened with "20 run reports consumed — full coverage, no gaps" while the directory it named held no files at all for that week. Its numbers were still correct, which is what made it dangerous.
An automated watcher reported "new bugs in sprint: 0" for the third day running. A Highest-priority bug had been raised three hours earlier, on a story outside the tracked set.