Claude Code

Your Opus 5.5 Prompt Checklist

Anthropic's own guide to prompting Opus 5.5 and Claude Code, boiled down to the five changes that matter, with the exact text to paste and the catch

2min read
1prompts to copy
5numbered steps
3sections

Opus 5.5 works longer on its own and thinks before every reply, so the old prompt habits can get in the way. This is the checklist from Anthropic's guide, in the order to try it.

The checklist

  1. Delete "think carefully" lines from your prompts. Opus 5.5 already thinks before every reply.
  2. Hand over the whole task in one message. Name the finish line, like "the tests pass" or "every endpoint is migrated", and say when it should stop and ask. Then let it work.
  3. In Claude Code, you can type a new message and press Enter while it works. Example from the guide: "Also keep the old endpoint names as aliases."
  4. Put stop rules in your CLAUDE.md (text below). Keep permission prompts on for destructive commands too.
  5. When a long run ends, read what it needs from you first.
PROMPT
PASTE INTO CLAUDE.md
When a step doesn't need my input, keep going. Put status notes in the same message as your next action.
Stop and ask only when you can't continue without me, or before anything destructive: deleting data, force-pushing, or changing anything outside this repository.

Keep reading

Free, for an email.

Unlocks 2 more sections, the document version, and every other guide on the site, for free. Enter your email once to keep reading.

45,000+ follow where these guides come from.No spam. Unsubscribe in one click.

Read next

Send Claude's Easy Jobs to Haiku 5.5 With One Helper File
We gave Opus and Haiku 5.5 the same job and both got every answer right. Haiku cost 96% less. The test, the one helper file that hands easy jobs to Haiku, the catch, and a prompt that sets it up
2 min
Anthropic's Three Sample Claude Code Mods, With the Setup Commands
A forecast of how full Claude's memory is, a guard that freezes rm -rf and shows what it would delete, and a replay of every edit Claude just made. What each one does, the exact commands to try them, and the prompt to have Claude build your own
4 min
Stop Hidden Orders in Your Files From Tricking Claude: Set Up the Free /trojan Detector in 2 Minutes
A file can hide a sentence in text you can't see, and when you hand that file to Claude, Claude can follow the hidden order and try to steal things like your passwords. Anthropic, the makers of Claude, call this kind of prompt injection one of the most significant security challenges for browser-based AI agents. This free Claude Code skill checks a file before Claude reads it: three checks, the hidden text shown in red. How to set it up, ready-made prompts, and what an answer looks like
11 min