core-ml

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes existing project code using Grep and Glob to determine how to integrate new features. While this allows the agent to ingest project-specific context, the interaction is restricted to developer-controlled source code and follows standard assistant patterns.
  • [DYNAMIC_EXECUTION]: The skill includes patterns for the runtime compilation and loading of Core ML models. This is a core feature of the framework used to optimize performance on specific Apple hardware (CPU, GPU, or Neural Engine) and is implemented with appropriate safety checks and error handling.
  • [EXTERNAL_DOWNLOADS]: Swift templates provided in the skill demonstrate how to securely download and update machine learning models from remote sources. These implementation examples use standard system APIs and include guidance on verifying model integrity and handling network availability.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 09:32 PM