The Agent Skills Directory

[SAFE]: The skill consistently uses the "defusedxml" library for all XML parsing operations in scripts/utilities.py, scripts/document.py, ooxml/scripts/unpack.py, and ooxml/scripts/pack.py. This effectively mitigates risks associated with XML External Entity (XXE) and other XML-based denial-of-service attacks.
[SAFE]: System command execution is handled securely through the use of "subprocess.run". Calls to external tools like LibreOffice (soffice) for document conversion and validation, and Git for redlining comparison, use hardcoded or temporary file paths, preventing arbitrary command injection.
[SAFE]: The skill operates within a temporary directory created via "tempfile.mkdtemp", ensuring that all document unpacking and manipulation are isolated from the rest of the file system and preventing unintended modification of original files unless explicitly saved.
[SAFE]: No hardcoded credentials, sensitive data access patterns, or data exfiltration mechanisms were detected in the scripts or prompts. The skill is authored by "iofficeai" and all resources are internal to the skill package.
[SAFE]: While the skill processes external document data (OOXML XML), which is a theoretical surface for indirect prompt injection, it follows the primary purpose of the skill and incorporates best practices for XML handling and sanitization.

docx