This skill automates the refinement of critical AI orchestration files such as CLAUDE.md, SKILL.md, and agent definitions. By employing a multi-phase workflow involving baseline metric measurement, specialized documentation agents, and independent verification, it ensures that your AI instructions are concise, instruction-rich, and formatted for peak LLM comprehension. It reduces token waste while significantly improving the effectiveness of agent behaviors through structural upgrades like decision flows, positive framing, and strategic XML tagging.
主要功能
01Automated token counting and 24-point skill completeness scoring
02Iterative optimization mode for large plugin directories and complex files
03Applied transformations based on 8 core LLM-optimization principles
0430 GitHub stars
05Multi-agent optimization workflow with independent quality verification
06Comprehensive before/after reporting with token delta and impact analysis
