The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/build.py script manages the conversion pipeline by invoking several external tools, including pandoc for document generation, node for building reference templates, and npm to resolve global package paths. These are executed using the subprocess module with arguments passed as lists to prevent shell injection.
[EXTERNAL_DOWNLOADS]: The scripts/render_mermaid.py utility performs network requests to the mermaid.ink service to convert Mermaid diagram code into PNG images. While this involves sending user-provided diagram code to an external service, it is a standard function of the tool and uses a well-known rendering API.
[DYNAMIC_EXECUTION]: The build process dynamically generates a Node.js script (build_ref.js) from a predefined template. This script is used to interface with the docx library to create a temporary reference document used by Pandoc during the conversion process.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied Markdown files, which serves as a surface for indirect prompt injection. 1. Ingestion points: Markdown files (input.md) and Mermaid code blocks. 2. Boundary markers: None identified; the skill directly processes the provided file content. 3. Capability inventory: The skill has the ability to read and write to the filesystem, execute subprocesses (pandoc, node), and perform network requests. 4. Sanitization: No specific sanitization or validation of the Markdown content or Mermaid code is performed prior to processing or transmission to the rendering API.

markdown2docx