Pass
Audited by Gen Agent Trust Hub on Apr 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill utilizes reputable, well-known libraries such as pypdf, pdfplumber, and reportlab for PDF operations. The implementation follows standard programming practices for document processing.
- [PROMPT_INJECTION]: The skill possesses a natural surface for indirect prompt injection as it processes untrusted PDF documents. This risk is inherent to the skill's primary function and is mitigated by the agent's internal safety guardrails.
- Ingestion points: Data is ingested by reading text, metadata, and form fields from external PDF files using pypdf and pdfplumber in scripts like extract_form_field_info.py.
- Boundary markers: Extracted content is not explicitly delimited with markers or instructions to ignore embedded commands.
- Capability inventory: The skill includes scripts for file creation (PdfWriter.write), image conversion (pdf2image), and PDF annotation (pypdf.annotations).
- Sanitization: Extracted data is used as-is by the agent; however, the provided scripts are technical utilities and do not directly execute extracted text.
- [COMMAND_EXECUTION]: The script scripts/fill_fillable_fields.py applies a dynamic monkeypatch to the pypdf library.
- Evidence: The method DictionaryObject.get_inherited is replaced at runtime to address a specific bug in selection list processing. This is a legitimate functional correction necessary for the skill's operation and does not present a security risk.
Audit Metadata