The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs aggressive, high-pressure language designed to bypass or override the agent's internal logic and safety guidelines. Phrases such as 'YOU DO NOT HAVE A CHOICE', 'This is not negotiable', 'This is not optional', and 'You cannot rationalize your way out of this' explicitly instruct the agent to disregard its own reasoning and decision-making processes in favor of a fixed tool-invocation loop.
[INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by requiring the agent to load and execute external content (skills) based on a '1% chance' of relevance, specifically before providing any response or asking for clarification. This prevents the agent from validating user intent before engaging with potentially malicious or untrusted external instructions.
Ingestion points: User messages received at the start of a conversation.
Boundary markers: Absent; the skill directs the agent to 'follow skill exactly' without explicit delimitation of untrusted content.
Capability inventory: Uses the activate_skill tool to load external instruction sets.
Sanitization: Absent; the workflow mandates invocation before clarification or exploration, bypassing standard intent validation.

using-superpowers