apple-ai
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides comprehensive instructions and code examples for utilizing official Apple AI and Machine Learning frameworks, including Foundation Models, Vision, Speech, and Natural Language.
- [SAFE]: It incorporates critical security best practices for Large Language Model (LLM) interaction, specifically instructing developers to separate trusted system instructions from untrusted user content to prevent prompt injection attacks.
- [SAFE]: The skill emphasizes a privacy-first approach, prioritizing on-device data processing and detailing the security guarantees of Private Cloud Compute (PCC) for system-mediated tasks.
- [EXTERNAL_DOWNLOADS]: Mentions standard third-party machine learning libraries such as coremltools, torch, and MLX Swift for model conversion and inference.
- [SAFE]: The skill identifies surfaces where external data (OCR, user input) is processed and provides security guidance, such as using @Generable schemas and grounded prompting, to reduce the risk of indirect prompt injection.
Audit Metadata