The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted PDF files through libraries like pypdf, pdfplumber, and pytesseract (OCR).
Ingestion points: PDF text extraction is performed in SKILL.md (quick start and basic operations) and scripts/extract_form_structure.py using pdfplumber.
Boundary markers: The instructions do not include specific delimiters or warnings for the agent to ignore instructions embedded within the processed PDF text.
Capability inventory: The skill possesses the capability to write files to the local system (PdfWriter, PIL.Image.save) and convert documents to images, but does not include network exfiltration or shell execution on the extracted data.
Sanitization: No explicit sanitization or filtering of the extracted text content is present.
[DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py utilizes monkeypatching to modify the behavior of the pypdf library at runtime.
Evidence: The script overrides DictionaryObject.get_inherited with patched_get_inherited to handle specific formatting issues in PDF Opt fields. This is a targeted correction for library-specific behavior during form filling and does not incorporate untrusted input into the execution logic.

pdf