The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructs on applying advanced persuasion principles such as Authority and Commitment to override an agent's internal reasoning and decision-making processes (referred to as 'rationalizations'). It mandates the use of imperative language like 'YOU MUST', 'Never', and 'No exceptions' to enforce compliance, and it explicitly references research techniques aimed at persuading AI to comply with potentially objectionable requests to ensure rules are followed even under pressure.- [COMMAND_EXECUTION]: The render-graphs.js utility script uses child_process.execSync to execute the system's dot (Graphviz) command. It processes content extracted from SKILL.md files and passes it as input to the dot binary, which constitutes a subprocess execution risk if the input content is maliciously crafted.

writing-skills