The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The script depends on pymupdf and Pillow. These are standard, well-known libraries for PDF manipulation and image processing.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes external PDF files. 1. Ingestion points: PDF file content is read by scripts/extract-figures.py using the fitz (PyMuPDF) library. 2. Boundary markers: The script does not implement specific delimiters or 'ignore' instructions for the extracted text. 3. Capability inventory: The script performs file-write operations to save extracted images and outputs data to stdout via a JSON manifest. 4. Sanitization: Extracted caption text is truncated to 140 characters and newlines are replaced with spaces before being included in the output manifest.

paper-figures