The Agent Skills Directory

Runtime Compilation and Library Injection: The skill includes a utility (scripts/office/soffice.py) that compiles a C-based shim at runtime using gcc and utilizes the LD_PRELOAD environment variable. This technical implementation is designed to shim Unix domain sockets, allowing LibreOffice to function correctly in restricted or sandboxed environments.
Dynamic Macro Execution: To automate document tasks, the skill (scripts/accept_changes.py) generates a LibreOffice Basic macro at runtime and executes it via the soffice binary. This is used to programmatically accept all tracked changes in a document.
External Command Execution: The skill orchestrates several external tools, including LibreOffice, pandoc, pdftoppm, git, and the gcc compiler. These tools are used for document conversion, content analysis, and environment compatibility.
Indirect Prompt Injection Surface: As the skill processes user-provided .docx files, it handles an indirect prompt injection surface. The implementation includes defensive measures, such as stripping symlinks from untrusted archives and performing rigorous XML schema validation to mitigate risks from untrusted content.
Secure File Handling: The skill employs best practices for file system operations, including a secure archive extraction routine (scripts/office/helpers/__init__.py) that mitigates path traversal (ZipSlip) risks and explicitly forbids symlink entries in archives.

docx