The Agent Skills Directory

[SAFE]: The skill's logic and instructions align with professional design workflows and do not exhibit malicious behavior or security risks.\n- [COMMAND_EXECUTION]: The skill uses local Bash commands (find, cat, sed) to detect its environment and locate related sub-skill files. These commands are used solely for internal path resolution and to load instruction sets from known local directories (e.g., plugin or user skill folders).\n- [REMOTE_CODE_EXECUTION]: A specific safety rule (Iron Rule 6) prohibits the agent from installing any software dependencies without the user's explicit permission, effectively preventing unauthorized remote code execution during the implementation phase.\n- [EXTERNAL_DOWNLOADS]: The skill references external Model Context Protocol (MCP) tools for visual asset and mockup generation. These are treated as optional integrations, and the skill itself does not perform any unauthorized downloads or suspicious network interactions.

design-excellence