Getting the most out of Opus 5.5 in Claude and Claude Code

When you use autonomous coding assistants like Claude Code, the way you write your prompts makes a huge difference. In a guide on the claude.dev blog, Addy Osmani shares that the best way to get great results on big tasks is to clearly define what "done" looks like right from the start. Instead of micromanaging each step, describe the whole project, explain the exact success criteria, and let the model work through the problem.

You can also ditch old prompting habits like asking the AI to "think step by step." Modern reasoning models already plan their steps before writing code, so extra prompting tricks just slow down your workflow. If you want faster answers on simple questions, you can simply tell the tool to answer directly.

For visual and frontend work, Osmani recommends listing specific design habits you want to avoid rather than asking for a generic vibe. Telling the tool to skip common defaults like cream backgrounds, italicized headings, or pill buttons forces it to produce fresher, more customized interfaces for your web apps.

Community discussion: https://news.ycombinator.com/item?id=49946567