skill-bloat-auditor-yashu

To audit skill documentation, scan SKILL.md and reference markdown files against a pattern library of bloat indicators to identify and report redundant, historical, or non-essential text. This ensures AI operational instructions remain concise and actionable.

Documentation bloat in skill development consists of redundant, outdated, or purely explanatory text within SKILL.md files. It includes historical notes, deprecated rules, and repetitive declarations that distract from active AI operational guidelines.

You need to clean up markdown documentation for AI agents after a major refactor or when maintenance drops. Auditing removes obsolete rules and explanatory text, ensuring only current, active instructions remain for the AI to follow.

An automated documentation audit categorizes findings into high-confidence bloat like obsolete or historical text, and medium-confidence bloat like principles or guides. This confidence-based reporting separates clear removals from items needing human review.

Yes, you can scan reference markdown files for deprecated rules. The auditor scans both SKILL.md and reference files against predefined bloat indicators to detect historical notes and repetitive self-contained declarations.

A limitation of automated documentation bloat detection is that medium-confidence findings, such as principles and guides, require human review to determine if they are truly non-essential before removal from the skill documentation.