image-to-text
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes local Python scripts (
ocr_words.py,zoom.py) that invoke thetesseractOCR binary via thesubprocess.runmethod. This execution is confined to local image processing and is a fundamental requirement for the skill's OCR functionality. - [INDIRECT_PROMPT_INJECTION]: The skill processes text extracted from external images, which represents an attack surface for indirect prompt injection. The instructions include robust mitigation strategies, such as the requirement for verbatim transcription ("Copy, do not improve"), mandatory verification against secondary routes (Vision vs. OCR), and arithmetic validation of extracted numerical data.
Audit Metadata