What struck me is how unglamorous most of these “habits” are. The advice is less “prompt better” and more “stop compensating for a model that no longer needs your crutches.” That feels right to me. A lot of Claude usage has become cargo cult: “think step by step,” “show your reasoning,” endless re-explanations of obvious constraints. If Opus 5.5 really does as much on its own as the post suggests, those habits are not clever anymore; they’re friction.
The strongest point here is the one about defining done. That’s not a Claude-specific trick, it’s just good delegation. But it matters more when the model can run longer and farther on its own. If you don’t pin down the finish line, you’re basically asking a contractor to wander around your kitchen until they feel spiritually complete. I’d absolutely try that style of prompt first: task, definition of done, and the one condition that should stop the run.
I’m less convinced by the more prescriptive bits around effort levels and subagents. Maybe they’re useful, but they also sound like the kind of guidance that ages fast as the model shifts. “Start at medium” is sensible, but really it’s just “measure before you spend.” Same with “ask for subagents by name”: probably a good forcing function for large codebases, though I suspect plenty of tasks will still be better served by a single careful run. I’d want to see where the break-even point actually is.
The screenshot advice is the most straightforwardly practical thing in the piece. If the model can read charts and diagrams better than Opus 5, then yes, stop retyping numbers into a prompt and risk introducing your own mistakes. That part feels like the least controversial upgrade here.
What I like overall is the shift in posture. The model is treated less like a junior who needs constant hand-holding and more like a worker who needs clear boundaries. That’s a healthier mental model. But I also think people will overread this and start deleting useful instructions just because “the model is smarter now.” Some of those old prompt habits were ugly, sure, but they existed for a reason. The real trick is knowing which ones were superstitions and which ones were doing actual coordination work.
