The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external PDF documents, which are untrusted sources. Text or data extracted from these files could contain instructions intended to mislead the AI agent. This is an inherent risk of document processing tools.
Ingestion points: Text extraction in SKILL.md (using pypdf or pdfplumber) and form metadata extraction in scripts/extract_form_field_info.py.
Boundary markers: No explicit markers are used to delimit extracted PDF content from agent instructions in the provided examples.
Capability inventory: The skill possesses file read/write capabilities (pypdf, pdfplumber, reportlab) and shell execution capabilities (via pdf2image and other utilities).
Sanitization: The skill performs specific data structure validation (e.g., in scripts/fill_fillable_fields.py) but does not sanitize PDF text content for potential injection attacks.
[DYNAMIC_EXECUTION]: The scripts/fill_fillable_fields.py script utilizes monkeypatching to modify the behavior of the pypdf library at runtime. Specifically, it overrides DictionaryObject.get_inherited to fix a known bug in the library's handling of selection lists. This is a legitimate and documented use case within the script's intended functionality.

pdf