Deconstructs Claude Code skills to separate deterministic automation from tasks requiring LLM judgment.
Skill Analyzer helps developers optimize their Claude Code workflows by auditing SKILL.md files and identifying steps that can be offloaded to scripts. It classifies every action as either mechanical (deterministic) or cognitive (LLM-based), providing a structured classification table, a proposed script interface for automation, and a leaner, more efficient version of the original skill definition. This ensures that the LLM focuses its reasoning power on high-value tasks while scripts handle predictable file I/O and CLI operations.
주요 기능
01Generates ready-to-use shell or Python scripts for deterministic tasks
02Classifies skill steps into scriptable and judgment-based categories
03Provides detailed rationales for every step classification
040 GitHub stars
05Refactors SKILL.md files to reduce complexity and improve reliability
06Summarizes automation potential and identifies workflow risks
사용 사례
01Refactoring complex manual skills into hybrid script and AI workflows
02Identifying brittle LLM steps that should be handled by regex or CLI tools
03Standardizing custom skills to lower token usage and improve performance
