The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes standard system utilities such as pandoc, libreoffice (soffice), and git to facilitate document conversion, validation, and word-level diffing. These operations are essential to the skill's primary purpose and are implemented using subprocess.run with safe argument lists, which prevents shell injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: As a document processor, the skill ingests user-provided Word documents that could contain malicious instructions. The skill proactively addresses technical risks associated with this by using defusedxml for all XML parsing operations, effectively mitigating XML External Entity (XXE) and recursive entity expansion attacks.
[EXTERNAL_DOWNLOADS]: The skill documentation correctly identifies and suggests the installation of reputable, standard dependencies required for its functionality, including pandoc, libreoffice, poppler-utils, and relevant Python/Node.js packages from established registries.
[SAFE]: Analysis of the 58 provided files reveals no evidence of malicious intent, hardcoded credentials, data exfiltration, or obfuscation. The skill's behavior align precisely with its stated purpose of document creation and editing.

docx