writing-skills

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • PROMPT_INJECTION (LOW): The files CLAUDE_MD_TESTING.md and persuasion-principles.md detail strategies to influence agent behavior through 'Authority' and 'Commitment' principles. These include imperative directives and behavioral overrides like 'YOU MUST', 'No exceptions', and failure-based framing designed to suppress the agent's natural reasoning or safety filters in favor of specific workflows.\n- COMMAND_EXECUTION (LOW): The render-graphs.js script executes the system binary dot via child_process.execSync. While it safely passes content from SKILL.md via standard input to prevent shell injection, the execution of local system binaries based on file content is a capability that should be monitored.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:49 PM