The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill relies on the 'requests' Python package for communication with the local API service, which is a standard and secure practice for this type of utility.
[COMMAND_EXECUTION]: The code demonstrates legitimate use of file system operations, such as reading input files and writing output Markdown and images, which are necessary for the skill's primary function.
[PROMPT_INJECTION]: The skill processes content from untrusted external documents, which constitutes a surface for indirect prompt injection. Ingestion points: Untrusted documents like .docx and .pdf files are read from the file system. Boundary markers: No specific boundary markers or 'ignore' instructions are applied to the converted content. Capability inventory: The skill has access to local file read/write operations and directory creation. Sanitization: The skill does not perform content sanitization on the text extracted from the source documents.

docling-convert