The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npm exec to install and run the kordoc package from the public npm registry. While the accompanying pdfjs-dist package is from a well-known source, kordoc is an external dependency from a third-party source not listed as a trusted vendor.
[DYNAMIC_EXECUTION]: The Node.js script dynamically loads code at runtime using import(). The target path (kUrl) is constructed using environment variables (process.env.PATH) to resolve the location of the installed kordoc package, which is a form of dynamic loading from computed paths.
[COMMAND_EXECUTION]: The skill executes shell commands including npm exec and node, passing a multi-line script via a heredoc to the Node.js interpreter.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to parse and extract metadata from HWP, HWPX, and PDF documents, creating an attack surface where malicious document metadata could influence agent behavior.
Ingestion points: Local document files provided or attached by the user (as specified in SKILL.md).
Boundary markers: No delimiters or safety instructions are used to wrap the extracted metadata before it is returned to the agent.
Capability inventory: The script has filesystem read access (fs.readFileSync) and the ability to execute commands via npm.
Sanitization: There is no evidence of sanitization or validation performed on the metadata extracted from the documents.

kordoc-parse-metadata