The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/office/soffice.py performs runtime compilation and process injection. It writes hardcoded C source code to a temporary file (lo_socket_shim.c), compiles it using gcc into a shared library, and then uses the LD_PRELOAD environment variable to inject this library into the soffice (LibreOffice) process. While documented as a Technical workaround for restricted environments, runtime compilation and library injection are highly sensitive behaviors.
[COMMAND_EXECUTION]: The script scripts/office/unpack.py is vulnerable to directory traversal (Zip Slip). It utilizes zipfile.extractall() without verifying that the extraction paths of the files inside the archive stay within the designated output directory. A maliciously crafted Word document could potentially overwrite sensitive system files or shell configuration files when processed.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting content from external Word documents. The instructions guide the agent to extract text using pandoc or scripts/office/unpack.py and bring it into the context without using specific boundary markers or sanitization logic to prevent the agent from obeying instructions hidden within the document's content.
[EXTERNAL_DOWNLOADS]: The skill instructions (SKILL.md) recommend the installation of the docx library from the public npm registry to enable document generation capabilities.

docx