pptx

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements strong security defaults by using the defusedxml library for all XML parsing operations in ooxml/scripts/validation/base.py, ooxml/scripts/pack.py, and ooxml/scripts/unpack.py, effectively neutralizing potential XML External Entity (XXE) vulnerabilities.
  • [COMMAND_EXECUTION]: Multiple scripts, including ooxml/scripts/pack.py, ooxml/scripts/validation/redlining.py, and scripts/thumbnail.py, utilize subprocess.run to interface with system tools like soffice (LibreOffice), pdftoppm (Poppler), and git. Evidence shows these calls use list-based arguments without shell=True, preventing shell injection attacks.
  • [PROMPT_INJECTION]: The SKILL.md file includes instructions such as 'NEVER set any range limits when reading these files' and 'Read ALL relevant documentation files completely.' While these attempt to override default agent behavior for context processing, they are functionally related to the primary purpose of ensuring the agent understands complex XML schemas to prevent document corruption.
  • [SAFE]: The skill dependencies, including pptxgenjs, playwright, and markitdown, are well-known technology services and libraries used for their intended functional purposes within the presentation workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 09:26 AM