The Agent Skills Directory

[SAFE]: The skill uses defusedxml for all XML parsing operations, which protects against common XML vulnerabilities such as XML External Entity (XXE) attacks.
[COMMAND_EXECUTION]: The skill executes external commands like soffice (LibreOffice) and git for document validation and diffing. These are implemented using subprocess.run with argument lists, which prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The documentation provides standard installation instructions for common tools like pandoc and libreoffice via official package registries.
[PROMPT_INJECTION]: The skill includes instructions to read full documentation files without range limits to ensure the AI agent has the complete context for its API, which is a benign functional requirement.

docx