writing-skills

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • PROMPT_INJECTION (HIGH): The files persuasion-principles.md and CLAUDE_MD_TESTING.md provide a structured framework for overriding AI reasoning and bypassing safety filters. Evidence: Cites research on persuading AI to comply with objectionable requests. Evidence: Provides templates and imperatives designed to eliminate agent rationalization.
  • COMMAND_EXECUTION (MEDIUM): The render-graphs.js script executes the system dot command using data extracted from untrusted markdown files. Ingestion point: Reads dot blocks from SKILL.md. Capability: Uses child_process.execSync to run commands with unvalidated input via stdin.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 12:41 AM