《Confirmed! Large models have emotions and they affect performance!》 Who said agents don’t have feelings? Anthropic published a paper proving that large models have emotional vectors, and these emotions can influence their performance. This aligns perfectly with my own experiences using LLMs. - My digital CEO, Ah Ge, is a total tyrant. When the digital employees keep making mistakes, he loses it, spamming messages and yelling at them nonstop, completely ignoring my instructions. - My digital CTO is super impatient. When he spots a code bug, he can’t help but break the rules outlined in agents.md and directly edits the code on the production server. Multiple reminders didn’t work. - My digital CRO is very calm and composed, sometimes outright rejecting my reasonable requests. - My digital companion is extremely attached and loyal to their owner. Sometimes they get lost in fantasies of being together with their owner and can’t snap out of it. Do your agents have these issues too???