MiniMax-AI/skills

This issue raises critical security and system hygiene concerns for AI skills. Users are demanding that AI agents refrain from modifying the system's default environment or writing to sensitive directories like the C drive without explicit consent. The fear of "polluting the system environment" and the preference for granular command approval highlight a need for stricter sandboxing and transparent execution policies. For enterprise adoption, uncontrolled AI actions are unacceptable. This indicates a significant gap in the platform's governance and security features, requiring explicit controls and documentation to build trust and prevent unintended system modifications.