The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute external tools including pandoc, soffice, git, and gcc for document processing and shim compilation.
[REMOTE_CODE_EXECUTION]: In scripts/office/soffice.py, the skill dynamically generates and compiles C source code at runtime using gcc. It then utilizes the LD_PRELOAD environment variable to inject the resulting shared library into the soffice process as a compatibility shim.
[PROMPT_INJECTION]: The skill processes untrusted document content, creating a surface for indirect prompt injection. Ingestion points: Word documents loaded for extraction or editing (e.g., via pandoc as referenced in SKILL.md). Boundary markers: No explicit delimiters or instruction-ignoring warnings are used to wrap extracted content. Capability inventory: Significant system control via shell execution and file system access. Sanitization: XML parsing is hardened with defusedxml, but document content text is not sanitized for potential instructions.

docx