The Agent Skills Directory

[PROMPT_INJECTION]: The skill extracts text content from external PowerPoint files, which creates a surface for indirect prompt injection where malicious instructions could be embedded in slides.\n
Ingestion points: Slide content is read via scripts/inventory.py and the markitdown package.\n
Boundary markers: No explicit boundary markers or warnings to ignore embedded commands are implemented during text extraction.\n
Capability inventory: The skill has access to system utilities (soffice, pdftoppm, git) and a headless browser (playwright) for slide rendering.\n
Sanitization: XML parsing is secured using defusedxml to prevent external entity injection (XXE) attacks.\n- [COMMAND_EXECUTION]: The skill invokes external command-line tools to support presentation processing and validation.\n
Evidence: ooxml/scripts/pack.py and scripts/thumbnail.py call soffice (LibreOffice) for PDF conversion and document validation.\n
Evidence: scripts/thumbnail.py calls pdftoppm to generate slide thumbnail images.\n
Evidence: ooxml/scripts/redlining.py executes git diff to compare and validate document revisions.

pptx