pdf-reader

Pass

Audited by Gen Agent Trust Hub on Apr 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (extract_pdf_text.py) using the uv tool to extract text from user-provided PDF files.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted PDF content.
  • Ingestion points: Content is read from external PDF files via the input_pdf path in extract_pdf_text.py.
  • Boundary markers: The script adds structural markers (e.g., ## Page N) but lacks explicit instructions or delimiters to warn the agent against following instructions embedded within the extracted text.
  • Capability inventory: The skill performs file read and write operations within the workspace; the extracted Markdown is intended for subsequent analysis by the agent.
  • Sanitization: Extracted text is normalized for formatting but not sanitized or filtered for potential malicious instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 26, 2026, 07:05 PM
Security Audit — agent-trust-hub — pdf-reader