The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill performs runtime compilation of system-level code.
The script scripts/soffice.py contains a C source string (_SHIM_SOURCE) which is written to a temporary file and compiled using gcc into a shared library. This library is used as an LD_PRELOAD shim to bypass Unix socket restrictions in certain environments.
scripts/accept_changes.py dynamically generates a LibreOffice StarBasic macro (ACCEPT_CHANGES_MACRO) and writes it to the user's LibreOffice profile to automate document changes.
[COMMAND_EXECUTION]: The skill relies heavily on executing external binaries and shell commands to perform its core functions.
scripts/soffice.py and scripts/accept_changes.py invoke soffice (LibreOffice) to convert files and run macros.
scripts/validators/redlining.py and scripts/validate.py invoke git and pandoc to perform diffs and text extraction.
The automated compilation in scripts/soffice.py invokes the gcc compiler.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface because it processes untrusted Word documents provided by users or external sources.
Ingestion points: Documents are unpacked and their XML content is loaded into memory in scripts/unpack.py, scripts/utilities.py, and scripts/document.py.
Boundary markers: There are no explicit instructions or delimiters used to separate extracted document text from agent instructions in the provided scripts.
Capability inventory: The skill has extensive file system access (read/write), shell command execution capabilities, and the ability to modify document structure.
Sanitization: The skill correctly uses the defusedxml library for all XML parsing, which effectively mitigates XML External Entity (XXE) and billion laughs attacks.

docx