AI/ML Integration
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements AI model integration using client-side ONNX and backend Replicate API services.
- Secret Management: It correctly uses environment variables (
process.env.REPLICATE_API_TOKEN) to handle authentication tokens, avoiding hardcoded credentials. - Trusted Sources: Model files are referenced from Hugging Face, which is a recognized and trusted service for machine learning assets.
- Standard Dependencies: It relies on well-known packages like
onnxruntime-webandreplicatefor its AI operations. - Code Safety: No malicious patterns such as obfuscation, remote code execution, or unauthorized data access were found in the implementation of image tiling, tensor preparation, or embedding caching.
Audit Metadata