The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local Python scripts such as doc_render.py and docx_repair.py to process documents. It also interacts with system utilities including docker for container management and ssh for remote guest access.
[REMOTE_CODE_EXECUTION]: The skill is designed to facilitate command execution on a remote Windows guest over SSH via the word-remote renderer. It manages the lifecycle of these remote environments using docker-compose and specialized shell scripts.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted document formats (e.g., .docx, .doc, .xlsx, .pptx) which serve as an attack surface for vulnerabilities in the underlying rendering engines like Microsoft Word or LibreOffice.
Ingestion points: The skill ingests external document files provided by the user for conversion.
Boundary markers: No specific boundary markers or instructions to ignore embedded instructions are implemented for the document content.
Capability inventory: The skill utilizes subprocess execution for local scripts, Docker management, and SSH-based remote rendering.
Sanitization: The skill includes a docx_repair.py script specifically designed to mitigate OOXML package corruption commonly found in Google Docs exports.

docx-render