Validates and standardizes Claude and Codex skills to ensure structural integrity and cross-flavor compatibility.
Skill Polisher is a meta-utility designed to maintain high standards across your agentic development environment by auditing .codex/skills and .claude/skills directories. It automates the verification of frontmatter, tool naming conventions, and structural consistency using deterministic scripts, allowing developers to perform integrity sweeps and generate standardized audit reports. Whether you are managing a single custom skill or a large library, this tool ensures your AI extensions remain functional, well-documented, and compatible across different AI frameworks.
主要功能
01Cross-flavor compatibility checks between different skill formats
020 GitHub stars
03Deterministic reporting with TASTE scores and issue summaries
04Automated structural integrity audits for Codex and Claude skills
05Helper scripts for local JSON audit outputs and mailbox integration
06Standardization of frontmatter and tool naming conventions
使用场景
01Auditing third-party skills for compatibility before integration into a project
02Running routine integrity sweeps to prevent skill degradation in complex repositories
03Enforcing organizational standards across a large library of custom AI skills
