pdf

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the pdftoppm command (part of the Poppler suite) to render PDF pages into images for visual review.
  • [PROMPT_INJECTION]: The skill processes untrusted PDF documents through text extraction and rendering, which creates an indirect prompt injection surface.
  • Ingestion points: External PDF files processed via pdfplumber and pypdf as described in SKILL.md.
  • Boundary markers: None identified; no specific delimiters or warnings to ignore embedded instructions are provided.
  • Capability inventory: Execution of system commands (pdftoppm) and Python code using the specified libraries.
  • Sanitization: No sanitization, escaping, or validation of the extracted PDF content is specified.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 12:39 PM
Security Audit — agent-trust-hub — pdf