docx
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands for document conversion and analysis, including pandoc for markdown extraction, libreoffice (soffice) for PDF generation, and poppler-utils (pdftoppm) for image conversion.
- [COMMAND_EXECUTION]: The skill facilitates the dynamic creation and execution of Python and JavaScript code to perform operations such as ZIP archive manipulation and OOXML structure editing.
- [COMMAND_EXECUTION]: The instructions suggest the use of sudo apt-get install to set up required system dependencies like pandoc, libreoffice, and poppler-utils.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of untrusted data from .docx files, which could contain instructions that attempt to hijack the agent's control flow.
- Ingestion points: Document content is extracted via pandoc and raw XML parsing of word/document.xml (SKILL.md).
- Boundary markers: The skill does not implement delimiters or system instructions to distinguish between document content and agent commands.
- Capability inventory: The agent has access to file writing, shell command execution, and package installation (SKILL.md).
- Sanitization: There are no mechanisms provided for escaping or validating extracted document text before it is presented to the agent.
Audit Metadata