Your CLAUDE. md is making Opus 5.5 DUMBER. Stuff like “double-check your work” and “think step by step” was advice for yesterday’s models. On Opus 5.5, it’s drag slower responses, weaker output. Lance Martin over at Anthropic just pushed the remedy. Open Claude Code and run: /claude-api prompt-audit It sweeps your skills, CLAUDE. md, and prompts. Strips the deadweight phrases that kneecap the model. Then rewrites what’s left using the official Opus 5.5 guidance. Takes five minutes. Do it tonight. Save this now, or you’ll swear you were going to later.
CyrilXBT@cyrilxbt197.6K followers6d ago
0:13Plays on X, where the post was published. Open the original post.
Post text
Your CLAUDE. md is making Opus 5.5 DUMBER. Stuff like “double-check your work” and “think step by step” was advice for yesterday’s models. On Opus 5.5, it’s drag slower responses, weaker output. Lance Martin over at Anthropic just pushed the remedy. Open Claude Code and run: /claude-api prompt-audit It sweeps your skills, CLAUDE. md, and prompts. Strips the deadweight phrases that kneecap the model. Then rewrites what’s left using the official Opus 5.5 guidance. Takes five minutes. Do it tonight. Save this now, or you’ll swear you were going to later.
This entry is part of the launch video directory and the AI Coding launch videos. Counts are read from X and refresh every 15 minutes; the video itself stays on X. See the methodology.