[Bug] DEFENSIVE CHECK EVERYWHERE IN Agent's JS CODE. NO WAY TO STOP #54
Description
GPT5 model in agent mode generates JS defensive checks every where, making the codebase unecessary ultra complicated & super hard to get the error to debug after each edit (the actual errors are all swallowed by TryCatch so it's super hard to track the code flow on unexpected behavior)
No way to stop it from doing so even Explicitly Banned defensive checks/ try...catch in 'copilot-instructions.md'
I need to tell it in the start of EVERY new chat session to respect the rule, but it violated again & again
Asked it why it violated the rule & it said that preventing runtime error was a higher innate priority that overwrite the rule
SO, I HIGHLY REQUEST A CONFIG ENTRY, A SETTINGS ENTRY TO STOP THIS BEHAVIOR FROM CO-PILOT!!!