The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts (pack.py, thumbnail.py, redlining.py) utilize subprocess.run to call system utilities such as soffice (LibreOffice), pdftoppm, and git. These calls are implemented securely using the list format for arguments, which avoids the use of a shell and mitigates potential command injection risks.
[DYNAMIC_EXECUTION]: The html2pptx.js library uses Playwright to render HTML into slides. It employs page.evaluate to execute its internal extraction logic within the browser context. This is a standard and necessary mechanism for the skill's high-fidelity conversion functionality.
[DATA_EXPOSURE]: The skill correctly identifies defusedxml as a dependency and uses it in several processing scripts to handle XML parsing, providing protection against XML External Entity (XXE) attacks. While some validation logic uses standard lxml, it operates on files that have already been handled by the unpacking routine.
[INDIRECT_PROMPT_INJECTION]: The skill extracts text from user-provided PowerPoint presentations via inventory.py and markitdown. While this creates an attack surface for indirect prompt injection, it is the primary intended purpose of the skill and the risk is mitigated by standard LLM guardrails.

pptx