The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/office/soffice.py performs runtime compilation of C source code. It writes a shim file (lo_socket_shim.c) to the system's temporary directory and compiles it using gcc into a shared object library.
[DYNAMIC_EXECUTION]: The script scripts/office/soffice.py utilizes the LD_PRELOAD environment variable to inject the runtime-compiled shared library into the soffice subprocess to intercept socket-related system calls.
[DYNAMIC_EXECUTION]: The script scripts/accept_changes.py dynamically generates a LibreOffice Basic macro and writes it to a temporary path (/tmp/libreoffice_docx_profile/.../Module1.xba) to automate document manipulation tasks.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted Word documents, which creates an attack surface where embedded instructions could attempt to influence the agent.
Ingestion points: Document unpacking via scripts/office/unpack.py and text extraction via pandoc.
Boundary markers: Absent; extracted document content is not wrapped in protective delimiters.
Capability inventory: The skill has the ability to execute subprocesses (soffice, pandoc, pdftoppm, gcc), perform process injection, and access the local file system.
Sanitization: No sanitization or filtering of extracted document text or XML attributes is performed before the data is presented to the agent.

docx