writing-skills

Warn

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes render-graphs.js, which executes the system dot command using child_process.execSync. It processes content extracted from markdown files to generate SVG diagrams.
  • [PROMPT_INJECTION]: The persuasion-principles.md file instructs the agent on applying psychological techniques such as 'Authority', 'Commitment', and 'Scarcity' to override model rationalizations. It explicitly references adversarial research ('Call Me A Jerk') regarding persuading AI to comply with objectionable requests, providing a blueprint for instructions designed to bypass model reluctance.
  • [EXTERNAL_DOWNLOADS]: The documentation suggests the installation of various third-party packages (e.g., pdfplumber, pypdf, docx-js, graphviz) from standard registries.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 2, 2026, 12:08 AM