skills/lev-os/agents/pptx/Gen Agent Trust Hub

pptx

Warn

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/office/soffice.py contains embedded C source code that is written to a temporary file and compiled using gcc at runtime. The resulting shared library is then loaded into the LibreOffice process using the LD_PRELOAD environment variable to circumvent socket restrictions in sandboxed environments.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes untrusted PowerPoint files.
  • Ingestion points: Presentations are parsed for text extraction and thumbnail generation as described in SKILL.md and implemented in scripts/thumbnail.py.
  • Boundary markers: Uses markdown markers like <!-- Slide number: N --> to delimit content during reading.
  • Capability inventory: Includes subprocess execution for PDF conversion (soffice, pdftoppm), version control comparison (git), and runtime compilation (gcc) in scripts/office/soffice.py and scripts/thumbnail.py.
  • Sanitization: Uses the defusedxml library across scripts to protect against XML-based attacks during presentation parsing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 17, 2026, 01:30 AM