image-ocr

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected in the skill instructions or code snippets. The skill serves as a library of best-practice implementations for OCR tasks.
  • [EXTERNAL_DOWNLOADS]: The skill recommends installing standard, well-known libraries from official registries (PyPI, NPM) such as pytesseract, easyocr, paddleocr, and official cloud SDKs from Google, AWS, and Anthropic. These are recognized as well-known technology providers.
  • [DATA_EXFILTRATION]: No hardcoded credentials or unauthorized data transmission patterns were found. The implementation examples demonstrate the use of environment variables for secure authentication with cloud providers, such as GOOGLE_APPLICATION_CREDENTIALS.
  • [PROMPT_INJECTION]: The skill correctly manages the attack surface for Indirect Prompt Injection by providing sanitization logic and guidance:
  • Ingestion points: External images and PDF files processed via multiple OCR engines in SKILL.md.
  • Boundary markers: Not explicitly defined in code templates, but the rules section advises treating OCR output as unvalidated text.
  • Capability inventory: Uses standard OCR libraries and cloud APIs for data extraction without granting excessive system privileges.
  • Sanitization: Includes a clean_ocr_text function employing regular expressions to remove non-printable characters, normalize whitespace, and correct common OCR misreads.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 03:02 AM