The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs high-pressure, imperative language designed to override the agent's default decision-making and safety reasoning. Instructions like 'ABSOLUTELY MUST', 'not negotiable', and 'not optional' are used to force the agent to invoke tools even before clarifying questions or gathering context. It explicitly labels standard agent behaviors (like seeking context) as 'rationalizations' or 'red flags' to be ignored, which mirrors techniques used in prompt injection to bypass model constraints.

using-superpowers