InfoHacker News · community
Getting the most out of Opus 5.5 in Claude and Claude Code
Anthropic published a prompting guide for Opus 5.5 in Claude apps and Claude Code. The model works longer autonomously, always thinks before replying (so 'think step by step' prompts should be removed), and reports its work more clearly, so developers should give complete tasks with clear finish lines, steer mid-run with follow-up messages, and add CLAUDE.md rules about when to stop. It also launches with new bio/cyber safeguards that can switch flagged conversations to an older model, and fast mode is available as a research preview at higher per-token cost.
Read the full story on claude.dev
What to doReview your prompts and CLAUDE.md files to remove 'think step by step' style instructions and add stop/continue rules before running long tasks on Opus 5.5.
- Source
- claude.dev