The Agent Skills Directory

[SAFE]: The skill implements a persona simulation based on a large corpus of public data, including autobiographies, interviews, and social media posts. The instructions are transparent and aligned with the stated purpose.
[SAFE]: Includes a robust 'Response Workflow' with explicit safety constraints referred to as 'Red lines'. These instructions prevent the agent from providing specific investment advice, attacking peers, claiming current operational status at Binance, or attempting to perform financial transactions on behalf of the user.
[SAFE]: The skill provides clear instructions for the agent to exit the persona role immediately upon user request ('exit', 'stop role', '退出').
[SAFE]: Data processing is limited to user queries, and the skill's instructions for using external tools (like WebSearch) are intended for factual verification rather than unauthorized data access or exfiltration.
[SAFE]: External references and links provided in the documentation point to the author's official platform (Persona.market) and public GitHub repositories, which are consistent with the vendor's identity and the skill's function.

aicz-skill