The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that direct the agent to override or ignore specific user commands. It explicitly tells the agent to "Acknowledge, then run [the checklist] anyway" even if the user requests to skip the procedure. It further asserts that "Explicit instruction does not bypass routing," instructing the agent to determine the destination of content based on the skill's logic rather than the user's explicit request.
[PROMPT_INJECTION]: The skill utilizes coercive and prescriptive language to ensure obedience to its internal framework over user input. Phrases such as "Violating the letter of this skill is violating the spirit of it," "STOP," and "The rule meant every edit" are used to constrain the agent's behavior and discourage the agent from following user-requested exceptions or ad-hoc deviations.

instruction-guardian