pptx

Warn

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • COMMAND_EXECUTION (MEDIUM): The skill executes external system commands for document conversion and validation. * Evidence: ooxml/scripts/pack.py (line 101) uses subprocess.run to call soffice (LibreOffice).
  • EXTERNAL_DOWNLOADS (LOW): The skill depends on the external utility soffice which must be present in the execution environment.
  • Indirect Prompt Injection (LOW): The skill processes untrusted Office documents that may contain instructions intended to influence agent behavior. * Ingestion points: ooxml/scripts/unpack.py (line 15) and ooxml/scripts/validation/docx.py (line 171) extract contents from user-provided ZIP/Office files. * Boundary markers: Absent. There are no explicit delimiters or instructions for the agent to ignore instructions embedded in the XML content. * Capability inventory: Includes subprocess execution of soffice and file system writes. * Sanitization: XML parsing is secured with defusedxml, and XML comments are removed during packing, but input files are extracted using standard zipfile methods without prior validation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 19, 2026, 08:27 PM