pdf-ocr-skill

Warn

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/pdf_ocr_processor.py contains an install_dependency function that uses subprocess.check_call to execute pip install commands. This allows the skill to modify the system environment by installing software at runtime.
  • [EXTERNAL_DOWNLOADS]: The skill attempts to download and install several Python packages (rapidocr_onnxruntime, rapid-doc, pymupdf, pillow) from external repositories if they are not already present on the system.
  • [DATA_EXFILTRATION]: When the 'siliconflow' engine is selected, the skill sends image data (converted to Base64) to an external API endpoint (api.siliconflow.cn). While this is consistent with the stated purpose of cloud-based OCR, it involves transmitting potentially sensitive document content to a third-party service.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 20, 2026, 10:15 PM