Foundation Models Reference FAQs

Question 1

When should I use this skill in Claude Code?

Accepted Answer

Use this skill when developing for iOS 26+, macOS 26+, or visionOS 26+. It is essential for implementing LanguageModelSession, creating structured outputs with @Generable macros, or migrating prototypes to production-ready on-device AI.

Question 2

What does the Foundation Models Reference skill do?

Accepted Answer

This skill equips Claude with a comprehensive technical library for Apple's Foundation Models framework. It provides instant access to API documentation, technical specifications for 3B parameter models, and implementation patterns for on-device AI.

Question 3

Does this skill include WWDC 2025 code examples?

Accepted Answer

Yes, it contains an extensive collection of code examples and implementation patterns aligned with WWDC 2025 sessions, covering everything from basic text generation to complex multi-turn interactions and tool protocols.

Question 4

How does this skill help with structured AI output?

Accepted Answer

It provides detailed references for the @Generable macro and @Guide constraints. This allows Claude to help you implement constrained decoding, ensuring the on-device model returns type-safe Swift objects instead of unpredictable strings.

Question 5

What technical model details are included?

Accepted Answer

The skill includes specs for Apple's 3B parameter, 2-bit quantized models, including context window limits (4096 tokens), performance optimizations, and specific use cases like summarization, entity detection, and classification.

Foundation Models Reference

Foundation Models Reference

주요 기능

사용 사례

주요 기능

사용 사례