writing-skills

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The provided utility render-graphs.js uses execFileSync to invoke the system's dot binary. This is used specifically to render Graphviz diagrams into SVG format for visualization by the user during skill development.
  • [EXTERNAL_DOWNLOADS]: The documentation (specifically anthropic-best-practices.md) references established Python libraries such as pypdf and pdfplumber. These are well-known, legitimate packages used for PDF processing tasks.
  • [PROMPT_INJECTION]: The skill includes documentation on how to use psychological principles like authority and commitment to improve an agent's adherence to rules. While these techniques use strong imperative language ("YOU MUST", "No exceptions"), they are used as a defensive mechanism to ensure quality and discipline in skill execution, rather than to bypass platform safety filters.
  • [DYNAMIC_EXECUTION]: The render-graphs.js utility extracts Graphviz code from markdown files and processes it through an external renderer. This behavior is a standard utility function intended for local file processing and is initiated by the user for visualization purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 02:06 PM
Security Audit — agent-trust-hub — writing-skills