Anthropic

Anthropic /context-engineering Most of it gone, no measurable loss. Anthropic says the cut was made for Claude 5 models, and its coding evaluations did not move. We removed over 80% of Claude Code's system prompt. - Anthropic, July 24 CLAUDE.md 1 2 3 never write comments, document appropriately, match the style surrounding code. Delete the strict rules for older, dumber models. Stop stuffing everything upfront. was to make Claude.md a giant handbook of every practice you have, find it Now the play is progressive keep Claude.md short. CLAUDE.md Keep your CLAUDE.md lightweight and briefly describe what your repo is for, but spend most of the tokens on gotchas inside of the codebase. For example, you may organize your code to keep types in one monolithic file and nowhere else. Avoid stating "the obvious things Claude should know by looking at your file system or your repo. Claude can't see from the code the detailed into separate only when it needs them. You don't have to by hand. Article Here's what we've learned about prompting this new class of models, and how you can utilize it to update your context engineering. We've put these best practices in 'claude doctor', use the command /doctor in Claude Code to rightsize your skills, and CLAUDE.md files. Unhobbling Claude Overall, we found that we were over-constraining Claude Code, both through our system prompt and in our CLAUDE.md files and skills. /doctor, built to your skills. And Claude.md files One command and it flags for you. like onboarding a senior engineer an intern. You tell them weird things codebase and trust them The teams from Claude 5 right now instructions,