What problem does it solve? Large agent-driven systems drift over time: agent definitions fall out of sync with actual skills, governance rules go unenforced, and cross-references break silently. This Skill performs a full consistency audit of the Improvement Loop system and produces a persistent, severity-classified report. ## Core Features & Use Cases - Agent & Skill Verification: Checks every agent definition for completeness, validates skill frontmatter, and confirms skill inventories match actual SKILL.md files. - Governance & Structure Compliance: Verifies governance documentation, fractal directory pattern completeness, and CLAUDE.md factual accuracy. - Cross-Reference Integrity: Greps for DD/IB references, markdown links, and handoff protocol paths to detect broken links. - Use Case: Before archiving a milestone, run a full audit to confirm the system is clean, then review the prioritized remediation table to schedule fixes. ## Quick Start Run a full system audit of the Improvement Loop and write the findings report to operations/system-audits.