The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py performs runtime monkeypatching by replacing pypdf.generic.DictionaryObject.get_inherited with a custom implementation. While intended to fix library behavior regarding PDF choice options, this technique modifies the execution environment dynamically.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted PDF files to extract text, tables, and form metadata, which can be used to influence agent behavior.
Ingestion points: PDF files are processed by scripts/check_fillable_fields.py, scripts/extract_form_structure.py, and scripts/convert_pdf_to_images.py.
Boundary markers: Absent. There are no instructions or delimiters defined to prevent the agent from following instructions embedded within the processed PDF content.
Capability inventory: The skill can write files, convert documents to images, and execute shell commands (magick, pdftotext, etc.).
Sanitization: scripts/fill_fillable_fields.py includes validation logic that checks provided values against allowed options for checkboxes, radio groups, and choice fields.
[COMMAND_EXECUTION]: The skill instructions in forms.md and SKILL.md direct the agent to execute various command-line tools including magick (or convert), pdftotext, qpdf, pdftk, and pdfimages for image manipulation and PDF structure analysis.
[EXTERNAL_DOWNLOADS]: The skill relies on several external Python libraries including pypdf, pdfplumber, reportlab, pytesseract, pdf2image, pandas, and pypdfium2, which are recommended for installation via pip. It also references JavaScript libraries pdf-lib and pdfjs-dist in its advanced documentation.

pdf