The Agent Skills Directory

[SAFE]: Cognitive framework instructions. The skill utilizes a "tree" mental model to guide the agent in drilling down into specific scenarios or abstracting common themes from user input. This is a benign reasoning enhancement.- [SAFE]: Scoped autonomous investigation. The instructions to "Investigate Independently" (reading code, invoking MCPs) are framed as a method to improve the quality of follow-up questions rather than taking unauthorized actions. The skill explicitly states that independent investigation is meant to ground questions more firmly and not to begin implementation.- [SAFE]: No malicious vectors detected. The skill does not include external downloads from unknown sources, hardcoded credentials, persistence mechanisms, or obfuscated content.

grilling