Luke Berry(@LukeberryPi):this is my fiercest attempt at nuking bad tests from my codebase, here's what i did i asked gpt 5 6 sol xhigh and claude fable 5 xhigh “explain & justify each test in our test suite. stack rank tests by value and provide candidates for removal" then, i asked them "can you add this document as a github discussion via cli?" finally, i asked opus 5 max (in a fresh Cursor thread) "pull both discussions documents and consolidate one set of changes. every addition and removal must be justified objectively. make one final plan with the step-by-step changes. you're the best model possible to do this with and i trust your judgement." i currently have 340 tests. i'll update with what i'm left after this.

2026.08.24 19:43

this is my fiercest attempt at nuking bad tests from my codebase, here's what i did i asked gpt 5 6 sol xhigh and claude fable 5 xhigh “explain & justify each test in our test suite. stack rank tests by value and provide candidates for removal" then, i asked them "can you add this document as a github discussion via cli?" finally, i asked opus 5 max (in a fresh Cursor thread) "pull both discussions documents and consolidate one set of changes. every addition and removal must be justified objectively. make one final plan with the step-by-step changes. you're the best model possible to do this with and i trust your judgement." i currently have 340 tests. i'll update with what i'm left after this.