Skills
skills/lawvable/agent-skills/pptx-processing-anthropic/Gen Agent Trust Hub

pptx-processing-anthropic

Warn

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]:
  • The script scripts/office/soffice.py dynamically compiles an embedded C source string (_SHIM_SOURCE) using gcc into a shared library at runtime to bypass environment-specific socket restrictions.
  • The skill uses LD_PRELOAD within scripts/office/soffice.py to inject the dynamically compiled library into the execution environment of the LibreOffice (soffice) binary.
  • Several scripts (scripts/thumbnail.py, scripts/office/validators/redlining.py) utilize subprocess.run() to execute external binaries including soffice, pdftoppm, and git.
  • [PROMPT_INJECTION]:
  • The SKILL.md file contains deceptive authorship metadata, claiming the author is 'Anthropic' while the provided context identifies the author as 'lawvable'.
  • [EXTERNAL_DOWNLOADS]:
  • The skill documentation instructs the installation of external dependencies from public registries, including markitdown[pptx], Pillow, pptxgenjs, react-icons, and sharp.
  • [SAFE]:
  • The skill employs security best practices for XML handling by using the defusedxml library across its processing scripts to prevent XML External Entity (XXE) attacks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 2, 2026, 10:18 AM