holy f*ck. make Grok Bot talk like a caveman and use around 65% fewer tokens for the same output
it cuts greetings, filler, and paragraphs leading up to the answer. you get the diagnosis and the fix immediately, while code, commands, and exact errors stay untouched
in the test:
normal explanation: 69 tokens
caveman version: 19 tokens
same diagnosis. same fix. less chatter to generate and read
copy & paste this prompt into Grok Bot
save this one
@rugikkk cat claimed a seat and the whole place complied actively honestly staying seated ignored details clever feeling satisfied usually awake eventfully quiet after
@Hedgeye was mostly good owning trash treats and always choosing differently old and my stupid lovely pet cat selected surprising perfectly, late times normal.