The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The Python script scripts/md2docx.py automatically downloads images from remote URLs specified in the input Markdown file using urllib.request.urlretrieve. This allows the tool to make arbitrary HTTP/HTTPS requests to external servers during the conversion process.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes untrusted Markdown data while possessing network and file system capabilities. Ingestion points: The script scripts/md2docx.py reads Markdown content from the file path provided via the --input argument. Boundary markers: No explicit delimiters or boundary instructions are used to separate the external content from the conversion logic. Capability inventory: The script scripts/md2docx.py performs network downloads (urllib.request.urlretrieve), file system writes (doc.save), and temporary file management (tempfile, os.unlink). Sanitization: The script does not validate or restrict the domains of remote images. Additionally, user-supplied watermark text is interpolated directly into a VML XML template in scripts/md2docx.py without escaping, which could allow for XML structure injection in the generated output document.
[SAFE]: The installation instructions in README.md and SKILL.md recommend using pip install --break-system-packages. While this flag bypasses default system-level Python environment protections, it is a standard approach for managing dependencies in development environments and does not indicate malicious behavior.

lov-any2docx