easyocr
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation notes that the easyocr library downloads pre-trained models from its official repositories upon first execution. This is a standard and expected feature of the library.
- [PROMPT_INJECTION]: The skill processes untrusted image inputs which could contain malicious instructions, creating a surface for indirect prompt injection.
- Ingestion points: Image objects passed to the reader in SKILL.md.
- Boundary markers: Not specified.
- Capability inventory: None detected within the skill content.
- Sanitization: The instructions include defensive measures such as filtering by confidence score and performing text normalization via regex_data_normalizer.
Audit Metadata