The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes several system binaries, including LibreOffice (soffice), Inkscape, Ghostscript, and ImageMagick, via Python's subprocess.run and Node.js spawnSync to handle document rendering and image conversion tasks.\n- [DYNAMIC_EXECUTION]: In the code.js utility, syntax highlighting modules from the prismjs library are loaded dynamically using language identifiers provided by the user, which could potentially be exploited to load arbitrary modules.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes complex external file formats such as PPTX and PDF. These files are processed by powerful system tools without explicit sanitization or boundary markers, creating a potential vector for indirect prompt injection if the files contain malicious instructions.\n
Ingestion points: render_slides.py (input_path), slides_test.py (input_path), ensure_raster_image.py (input_files).\n
Boundary markers: Absent.\n
Capability inventory: Shell command execution via subprocess.run (soffice, inkscape, gs, JxrDecApp, heif-convert, magick) in Python scripts and spawnSync (fc-match) in text.js.\n
Sanitization: Absent; file paths and content are processed directly by system utilities.

slides