The Agent Skills Directory

[COMMAND_EXECUTION]: The ooxml/scripts/unpack.py utility is vulnerable to a Zip Slip attack because it uses the zipfile.extractall() method without validating that the file paths within the presentation archive are safe. A maliciously crafted presentation could use directory traversal sequences (e.g., ../../) to overwrite critical system files when the AI agent attempts to unpack it.\n- [PROMPT_INJECTION]: The skill facilitates the analysis of external PowerPoint files, including slide text, speaker notes, and comments, without sanitizing the input. This creates an attack surface for indirect prompt injection, where an attacker could embed hidden instructions in a presentation to manipulate the agent's behavior during analysis.\n- [COMMAND_EXECUTION]: Multiple scripts (ooxml/scripts/pack.py, scripts/thumbnail.py, ooxml/scripts/validation/redlining.py) execute external system binaries like soffice (LibreOffice), pdftoppm (Poppler), and git via subprocess.run(). While these are used for legitimate document processing tasks, executing system-level binaries on untrusted input files presents an inherent risk of exploiting vulnerabilities in the underlying system software.

pptx