The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill processes untrusted Office documents. While it uses defusedxml to prevent XXE, the resulting text could influence agent behavior. \n- Ingestion points: ooxml/scripts/unpack.py and ooxml/scripts/validate.py. \n- Boundary markers: Absent. \n- Capability inventory: File system access, zip manipulation, and soffice execution. \n- Sanitization: Employs defusedxml to mitigate XML-based vulnerabilities. \n- Command Execution (SAFE): Uses subprocess.run with argument lists for LibreOffice validation, avoiding shell injection risks.

docx