pptx

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION] (LOW): The pack.py script executes the soffice (LibreOffice) command locally to validate documents via conversion. This is a standard use of a local utility with arguments derived from provided file paths.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill processes untrusted OOXML files. It implements security best practices by using the defusedxml library in pack.py and unpack.py to prevent XML-based attacks such as XML External Entity (XXE) injection and Billion Laughs expansion bombs.
  • [DATA_EXFILTRATION] (SAFE): No network operations, data exfiltration patterns, or connections to external domains were identified. All processing occurs on the local filesystem.
  • [REMOTE_CODE_EXECUTION] (SAFE): No patterns for remote code execution, dynamic code evaluation (e.g., eval/exec), or downloading of external scripts were found.
  • [CREDENTIALS_UNSAFE] (SAFE): No hardcoded credentials, API keys, or secrets were detected in the source code or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:50 PM