The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run within pack.py and redlining.py to call soffice (LibreOffice) and git. These tools are used for the primary functional purpose of document conversion and validation. The execution uses list-based arguments with shell processing disabled, which prevents command injection.
[EXTERNAL_DOWNLOADS]: The documentation references standard external dependencies required for document processing, including pandoc, libreoffice, poppler-utils, and the docx NPM package. These are well-known tools from trusted sources.
[PROMPT_INJECTION]: Forceful instructional language such as "MANDATORY
READ ENTIRE FILE" is used in SKILL.md to guide the AI agent's processing of library documentation. These are benign instructions intended to prevent truncation of context during API learning.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it ingests untrusted .docx files. While it lacks explicit natural language boundary markers for extracted text, it employs high-quality technical sanitization by using the defusedxml library for all XML parsing, effectively preventing XXE and other XML-based exploits.

docx