Anthropic cut over 80% of Claude Code's system prompt for the Claude 5 generation — and says nothing measurable broke

On 24 July 2026 Anthropic published new context-engineering guidance for Claude 5-generation models, reporting it removed more than 80% of Claude Code's system prompt for Opus 5 and Fable 5 "with no measurable loss on our coding evaluations." The guidance trades hard rules for judgment — the old "never write multi-paragraph docstrings" instruction became "write code that reads like the surrounding code: match its comment density, naming, and idiom" — drops tool-use examples in favour of expressive tool parameters, moves detail into selectively-loaded skills instead of front-loading it, and tells builders to stop hand-saving context that automatic memory already carries.

What it means Most agent harnesses still run the previous generation's playbook — long rule lists, worked examples, everything front-loaded. If the vendor's own harness got dramatically shorter without getting worse, the prompt you are maintaining is probably carrying dead weight.