The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides tools and patterns for extracting text and form data from PDF documents (e.g., document.string, page.string, annotation.widgetStringValue). This data constitutes untrusted input that could contain malicious instructions intended to bypass agent safeguards.
Ingestion points: PDFDocument(url:) and PDFDocument(data:) methods for loading external files are described in SKILL.md and references/core-implementation.md. Data extraction methods like extractFormFields are detailed in references/pdf-generation-and-forms.md.
Boundary markers: The instructions do not define delimiters or specific warnings to ignore instructions embedded within the PDF content.
Capability inventory: The skill includes capabilities to write to the local file system via document.write(to:) and open external URLs via UIApplication.shared.open(url) inside the ManagedPDFView coordinator in references/pdf-viewing-and-annotations.md.
Sanitization: There is no evidence of sanitization, validation, or filtering applied to extracted PDF content before it is presented to the agent context.

pdfkit