The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The ooxml/scripts/pack.py script executes the soffice system command for document validation. This is performed using subprocess.run with a list of arguments, which prevents shell injection vulnerabilities.
[REMOTE_CODE_EXECUTION] (SAFE): No remote code execution patterns or unauthorized package installations were detected. The skill uses standard Python libraries and known dependencies like defusedxml and lxml.
[PROMPT_INJECTION] (LOW): As the skill is designed to unpack and display XML content from Office documents, it presents an indirect prompt injection surface.
Ingestion points: The unpack.py and docx.py scripts read XML data from user-provided .docx, .pptx, and .xlsx files.
Boundary markers: None explicitly present in the provided templates to isolate document content from instructions.
Capability inventory: The skill can read/write files and execute the soffice command.
Sanitization: The skill uses defusedxml.minidom which provides protection against common XML vulnerabilities (XXE), but does not sanitize the semantic content for LLM instructions.

docx