The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The script ooxml/scripts/pack.py uses subprocess.run to call soffice (LibreOffice) for headless document validation. The arguments are passed as a list, and the operation is restricted to validating the processed document's integrity.
[DATA_EXFILTRATION] (SAFE): No network communication or hardcoded credentials were found. File operations are confined to the user-specified directories and temporary paths.
[REMOTE_CODE_EXECUTION] (SAFE): The skill does not download or execute remote scripts. It relies on standard libraries and trusted packages for its functionality.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill possesses an attack surface for indirect prompt injection as it processes external Office files. However, it implements standard mitigations.
Ingestion points: ZIP extraction in ooxml/scripts/unpack.py and XML parsing in validation modules.
Boundary markers: None (standard for file processing tools).
Capability inventory: File system access (read/write), ZIP extraction, and subprocess execution for soffice.
Sanitization: The implementation uses defusedxml.minidom and lxml.etree for secure XML handling, effectively preventing XML External Entity (XXE) attacks.

pptx