якщо вас теж задовбало шо Claude Code постійно недороблює роботу до кінця, то от вам скрипт для патчингу апки, який змінює зашиті промптина простих задачках різниці не побачите, але на чомусь більш складному змінена версія точно покаже себе краще, тест на скріншотіhttps://gist.github.com/roman01la/483d1db15043018096ac3babf5688881Patch: Output efficiency IMPORTANTBefore: "Try the simplest approach first. Do not overdo it. Be extra concise."After: "Choose the approach that correctly and completely solves the problem."Patch: Brevity ≠ thoroughnessBefore: "Lead with the answer, not the reasoning"After: Adds: "these guidelines apply to messages, NOT to thoroughness of code changes"Patch: One sentence ruleBefore: "If you can say it in one sentence, don't use three"After: "does not apply to the thoroughness of your implementation work"Patch: Anti-gold-platingBefore: "Don't add features beyond what was asked"After: "if adjacent code is broken or contributes to the problem, fix it"Patch: Error handlingBefore: "Don't add error handling for scenarios that can't happen"After: "Add error handling at real boundaries where failures can occur"Patch: Three lines ruleBefore: "Three similar lines better than premature abstraction"After: "Use judgment about when to extract shared logic"Patch: Subagent gold-plate (2x)Before: "don't gold-plate, but don't leave it half-done"After: "Do the work a careful senior developer would do, including edge cases"