Hacker News new | ask | show | jobs
by niobe 56 days ago
Having to "unprompt" behaviour I want that Anthropic thinks I don't want is getting out of hand. My system prompts always try to get Claude to clarify _more_.