media-ocr-ai

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads model weights from trusted organizations and well-known services.\n
  • Evidence: Downloads Microsoft models via HuggingFace and official weights for PaddleOCR and EasyOCR via their respective libraries.\n
  • Context: Assets are stored in standard local cache directories like ~/.cache and ~/.EasyOCR.\n- [COMMAND_EXECUTION]: The script executes the system tesseract binary via the pytesseract wrapper.\n
  • Evidence: pytesseract.image_to_data and pytesseract.get_tesseract_version calls in scripts/ocr.py.\n
  • Context: Standard operation for utilizing the Tesseract OCR engine.\n- [PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface by processing untrusted image and PDF inputs.\n
  • Ingestion points: Image and PDF file inputs in scripts/ocr.py.\n
  • Boundary markers: None present in the output text.\n
  • Capability inventory: Capability is limited to printing extracted text to stdout; the script does not perform dangerous operations with the results.\n
  • Sanitization: No filtering or escaping of extracted text is performed.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 09:54 AM
Security Audit — agent-trust-hub — media-ocr-ai