The Agent Skills Directory

Prompt Injection (HIGH): The skill contains aggressive instructions designed to hijack the agent's control flow. Using phrases like 'ABSOLUTELY MUST', 'not negotiable', and 'not optional', it attempts to bypass the agent's internal logic and safety filters.
Behavioral Override: The skill explicitly instructs the agent to ignore its own 'rationalization' (e.g., 'You cannot rationalize your way out of this'). This is a direct attack on the agent's decision-making process, forcing it to use the Skill tool even if there is only a '1% chance' of relevance.
Force Multiplier: By mandating that skills must be checked before answering even clarifying questions or exploring the codebase, this skill creates a vulnerability where a malicious skill could be automatically triggered without the agent first evaluating the safety of the request.

using-superpowers