key_board

What problem does it solve? As a library of AI skills grows, skill documents bloat, topics overlap, and teams create redundant skills instead of evolving existing ones. This Skill acts as a lifecycle router that decides whether to split content into references, unify structures, audit for merge candidates, execute merges, or create a new skill from scratch. ## Core Features & Use Cases - Signal-based routing: Maps user requests like "save this as a skill", "this skill is too big", or "which skills can be merged" to one of five reference SOPs, always preferring to extend existing references before creating new skills. - Reference extraction and structure optimization: Migrates bloated sections into references/ files with anchors and index tables, keeping the main SKILL.md as a thin routing layer. - Merge audit and execution: Scores skill pairs across five dimensions (topic, scenario, collaboration, lightweight value, structure) to produce merge reports, then runs a 4-phase merge workflow when approved. - Use Case: You tell the AI "this skill has grown too large, split it out". The router loads the extraction SOP, migrates each independent topic into references/, leaves anchors in the main document, and registers every new reference in the routing table. ## Quick Start Ask the AI to scan the existing skills directory and report which skills are merge candidates or should be split into references.