What problem does it solve? AI rules files like CLAUDE.md, AGENTS.md, or .cursor/rules/*.mdc often grow bloated, vague, and poorly ordered, which wastes context window and causes agents to ignore critical instructions. This Skill audits any rules-for-AI markdown file and returns a structured scorecard with actionable fixes. ## Core Features & Use Cases - Five-check audit: Scores length, direct code snippets, vague language, redundant knowledge, and rule ordering with OK/WARN/FAIL verdicts. - Grounded rewrites: Translates vague phrases like "write clean code" into testable rules based on the project's actual stack and conventions. - Interactive reordering: Proposes a better section order based on U-shaped attention and, only with explicit approval, reorders the file while preserving all content. - Use Case: A developer suspects their 600-line CLAUDE.md is being ignored by the agent. They run the review, get a scorecard showing 4 vague phrases and critical security rules buried at line 312, then approve a reorder that moves hard rules to the top. ## Quick Start Ask the assistant to review your AI rules file, for example: run the 10x rule review on @CLAUDE.md and show me the scorecard.
