The Agent Skills Directory

[SAFE]: The skill consists entirely of natural language instructions. It provides a structured workflow for turning ideas into designs through a series of gated steps (Understanding Lock, Design Exploration, Decision Logs).
[DATA_EXFILTRATION]: While the skill instructions direct the agent to review project files and documentation to establish context, this is a legitimate function of a design tool. There are no patterns suggesting data exfiltration or unauthorized access to sensitive system files.
[PROMPT_INJECTION]: The skill uses instructional language to constrain the agent's behavior (e.g., 'not allowed to implement', 'hard gate'), which are standard procedural constraints and do not attempt to bypass system-level safety guidelines.

brainstorming