rule-review

What problem does it solve?

This Skill helps you verify that an agent-rules file is correct, efficient, non-redundant, and actually reachable—so your agent follows the rules instead of ignoring or contradicting them.

Core Features & Use Cases

• Evidence-based auditing across seven dimensions: checks length, embedded code/config blocks, language precision, redundancy with your auto-active rules layer, ordering, cross-tool drift, and dead rules using grep.
• Cross-tool drift detection: compares common sibling rule sources (e.g., AGENTS.md vs CLAUDE.md vs GitHub Copilot vs Cursor) to find contradictions for the same scope.
• Safe remediation with --fix: applies only non-destructive, safe rewrites (e.g., removing redundancy, reordering critical sections) and flags anything that requires manual attention.

Quick Start

Run a read-only audit of your rules file by instructing the AI:
“Review AGENTS.md for redundancy, drift, ordering, and dead rules.”