The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a local Python script scripts/convert_word_to_md.py to perform document conversion. This script includes a routine that deletes the destination directory using shutil.rmtree if it already exists, which could lead to data loss if an incorrect path is provided.
[COMMAND_EXECUTION]: The conversion script uses the standard xml.etree.ElementTree module to parse XML content within .docx files. This module is not secure against maliciously constructed XML data (XXE), though the risk is localized to the parsing of the document structure.
[EXTERNAL_DOWNLOADS]: The documentation (references/setup.md) provides instructions to download and install Python and the markitdown package. These resources are fetched from official and well-known services such as Microsoft's GitHub/PyPI presence and official system package managers (winget, brew, apt-get).
[PROMPT_INJECTION]: The skill processes external, user-provided Word documents and converts them to Markdown for agent analysis, which is a vector for indirect prompt injection.
Ingestion points: The skill reads .docx files through the scripts/convert_word_to_md.py script.
Boundary markers: None. The agent is instructed to read the output Markdown directly without specific delimiters or instructions to ignore embedded commands.
Capability inventory: The conversion script has local file system read/write access and the ability to delete directories.
Sanitization: None. The script focuses on format conversion and does not filter the text content for potential injection attacks.

convert-word-to-md