smart-ocr

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides legitimate domain knowledge and code examples for text extraction tasks.
  • No malicious patterns such as prompt injection, credential exposure, or unauthorized persistence were detected.
  • All described behaviors align with the skill's stated purpose.
  • [EXTERNAL_DOWNLOADS]: The skill includes code snippets for retrieving image data from external URLs using the requests library.
  • Evidence: Usage of requests.get() to fetch image content for OCR processing.
  • Analysis: The targets are well-known placeholder domains used for demonstration, representing standard functional requirements for the task.
  • [COMMAND_EXECUTION]: File system operations are restricted to reading input files and managing temporary image pages during PDF processing.
  • Evidence: Use of os.remove() to clean up temporary files and pdf2image for page conversion.
  • Analysis: These operations are performed using standard libraries and do not involve execution of untrusted commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 08:21 AM