skills/kjgarza/marketplace-claude/pdf/Gen Agent Trust Hub

pdf

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns detected. The skill uses standard PDF processing libraries (pypdf, pdfplumber, reportlab) and provides helper scripts for common tasks like OCR and form validation.
  • [DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py implements a runtime monkeypatch of the pypdf library's DictionaryObject.get_inherited method. This is a localized modification used to resolve a specific library bug regarding selection list formatting and does not involve executing untrusted code or external data.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PDF files and agent-generated JSON metadata, creating an indirect prompt injection surface.
  • Ingestion points: PDF files are read in scripts/extract_form_field_info.py and scripts/fill_fillable_fields.py. Visual analysis instructions in forms.md rely on the agent interpreting converted images.
  • Boundary markers: None explicitly defined for content within the PDFs.
  • Capability inventory: The skill can write files and manipulate PDF structure using pypdf and Pillow.
  • Sanitization: The skill includes several validation scripts (scripts/check_bounding_boxes.py, scripts/fill_fillable_fields.py) to verify coordinate integrity and field values before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 11:01 PM