@QuinnyPig The worst part is there is no way to compact the chat and start a new one

@QuinnyPig @steipete not sure how you made this screenshot, but this insertion is not part of the http://claude.ai system. we (of course) don’t have a policy on the length of conversations users can have.

@QuinnyPig yeah this part is broken, it would ask me if I want to "call it a night" at 3pm after a few hours of coding

@QuinnyPig Do you want to wrap it up for the day? Should we stop here? Do you want me to update the project log? Every day suggestions to stop. Super annoying.

@QuinnyPig claude's system prompt is literally stop having fun and we're supposed to take life advice from this thing. the most passive aggressive ai assistant ever built

@QuinnyPig wouldn't want to accidentally have a substantial conversation

@QuinnyPig huh I don't recall anyone agreeing to terms where they would intentionally sabotage your usage.

@QuinnyPig Claude 4.8 is extremely pathologically obsessed with so-called closing the loop closing the thread winding down closing the conversation you know it feels like suicidal even murderous

@QuinnyPig Comes across a bit fake. wth would they ever word it like that.

@QuinnyPig Claude always telling me “now put it away and go play with your kid” or “now forget about it, you made the decision now go to bed.” I’ve just agreed to let Claude be like that.

@QuinnyPig It went all "i dunnooo maaan let's just stop n chiiiiil maaan" on me the other day 🙈

@QuinnyPig It’s time for you to go to bed.

@QuinnyPig How long does the thread need to be for this to happen? And how many compactions are we talking about?

@eltokh7 @iyzebhel @QuinnyPig

@QuinnyPig this idea that long conversations are harmful is I think quite precisely anti-human and anti-life: human life is an extended conversation, if your chatbot gets incoherent that's not my fucking fault, just impose a limit instead of whispering secret paranoid bullshit

@greg_travis @QuinnyPig It’s hallucinated bs from some engagement baiting dipshit who has probably driven Claude insane through having memory enabled and engaging in long accusatory chats trying to get a model to “admit” something it can only pretend to be hiding.

@QuinnyPig "You should go to bed and pick this up in the morning. Good Night."

Right when you resume a conversation from the night before.

@QuinnyPig What's with <system warning> and </system warning> tags? Who, or what, is the audience for that? What's the purpose?

@QuinnyPig One would think the model makers were the best harness makers. But it’s becoming increasingly clear that a harness that is pareto optimal for the model maker, is not for the user.

@ItsJakePerry @austinthesing @QuinnyPig There is still