The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs coercive and non-negotiable directives designed to override the agent's internal reasoning and decision-making logic. It uses absolute markers such as "ABSOLUTELY MUST", "not negotiable", and "not optional" to enforce compliance with a specific workflow, regardless of the agent's own evaluation of the situation.
Evidence: The skill includes a block tagged with <EXTREMELY-IMPORTANT> that explicitly tells the agent, "You cannot rationalize your way out of this." Furthermore, the "Red Flags" table instructs the agent to suppress its own critical reasoning, stating "These thoughts mean STOP—you're rationalizing," effectively attempting to disable the agent's safety and logic filters.
[INDIRECT_PROMPT_INJECTION]: The skill mandates that every incoming user message be used as a trigger to load external content via the Skill tool, creating an automated pipeline where untrusted user input can influence the execution of additional instructions or tools.
Ingestion points: Every user message received by the agent is processed as a trigger for skill invocation, as documented in the skill_flow diagram in SKILL.md.
Boundary markers: The skill lacks any instructions for the agent to sanitize user input or to differentiate between legitimate user requests and malicious instructions embedded in the input.
Capability inventory: The skill requires the use of the Skill tool to load content and the TodoWrite tool to generate tasks based on the invoked content.
Sanitization: There is no evidence of sanitization, validation, or escaping of the user-provided data before it is used to determine which skills to load and follow.

using-supaconductor