The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/office/soffice.py contains embedded C source code that is written to a temporary file and compiled using gcc at runtime. The resulting library is then injected into the LibreOffice process via the LD_PRELOAD environment variable to hijack low-level system calls such as socket, listen, and accept. Furthermore, scripts/accept_changes.py dynamically writes a StarBasic macro to the filesystem and executes it within the LibreOffice environment.
[COMMAND_EXECUTION]: The skill uses subprocess.run to invoke the gcc compiler at runtime in scripts/office/soffice.py. It also executes other system binaries including soffice, pandoc, pdftoppm, and git across multiple scripts to perform document conversions and comparisons.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to unpack and process XML content from untrusted .docx files, presenting a vulnerability surface where instructions embedded in a document could influence the agent.
Ingestion points: Word document XML content is processed by scripts/office/unpack.py and scripts/document.py.
Boundary markers: No delimiters or markers are used to separate document content from agent instructions.
Capability inventory: The skill has extensive command execution and process manipulation capabilities.
Sanitization: No input validation or sanitization is performed on text extracted from documents.

docx