💻 Tell Claude's new model "never" and it won't budge

💻 Tell Claude's new model "never" and it won't budge. Anthropic says what to write instead Claude Fable 5.2 treats any instruction with "never" as absolute. Its prompting guide suggests "ask for approval before deleting files" over "never delete files" when you still want a say. Newsletter writer Ruben Hassid turned the guide into lines anyone can paste: 1. Find "always" and "never" in your Claude instructions and memory. In grey areas, write "ask me first". 2. Don't just say "Go". Say what you approve: "Yes, rewrite the 3 welcome emails. Leave the invoice template alone." 3. Tell it to finish: Finish the whole job before you stop. Don't offer to continue: continue. Ask me before you delete anything. 4. Ask for a report you can read: End with a recap for someone who saw none of the work: key points first, plain words. In Anthropic's replay of real coding sessions, its longer finish-the-job text roughly halved the runs that stopped with work left. @aipost 🏴

