Intel / article

The New Rules of Context Engineering for Claude 5 Generation Models

Source: claude.com
Author: trq212
Date: Published Jul 24, 2026

Context

Instructions retained from older models can make a coding work harder to resolve contradictions. Anthropic reports removing more than 80% of Claude Code's for models including Opus 5 and Fable 5 without a measurable decline on its coding evaluations. Its account describes replacing blanket rules with guidance to follow the surrounding code, and moving specialized review instructions into skills loaded when needed.

The change concerns the whole bundle of information an agent receives, including project guidance, tools and memory. Anthropic's earlier context-engineering article (original) described loading relevant material as work proceeds, and cautioned that a minimal prompt must still give the agent enough information up front.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
system prompt — The standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.

Key quotes

“We removed over 80% of Claude Code’s system prompt for models like Claude Opus 5 and Claude Fable 5 with no measurable loss on our coding evaluations.”
trq212

“Overall, we found that we were overconstraining Claude Code, both through our system prompt and in our CLAUDE.md files and skills.”
trq212

“The number one rule for tool usage was to give Claude examples on how to use them. With our newest models, we’ve found that giving examples actually constrains them to a certain exploration space.”
trq212

“Some of our tools are ‘deferred loading,’ which means the agent must search for their full definitions using ToolSearch before using them. This allows us to have more tools (such as our Task tools) that don’t take up context until they’re needed.”
trq212

“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.”
trq212

Read the source claude.com