Can I use this for real-time UI updates?

Yes, it includes patterns for Snapshot Streaming, allowing you to display partially generated content in SwiftUI views as the model processes the request.

What does the FoundationModels Claude Code skill do?

It provides Claude with specialized knowledge to help developers implement Apple's on-device language models, covering session management, structured data extraction, and tool calling.

What is the benefit of using @Generable with this skill?

@Generable allows the on-device model to return structured Swift types instead of raw strings, providing compile-time safety and easier integration with app data models.

Does this skill help with offline AI implementation?

Yes, it focuses specifically on the FoundationModels framework which runs locally on Apple devices, ensuring AI features work without an internet connection.

How does this skill handle the 4,096 token limit?

The skill includes best practices for chunking large inputs and managing context windows within the hardware constraints of on-device execution.

FoundationModels On-Device LLM

Name: FoundationModels On-Device LLM
Author: saar120

bysaar120

0•

Mobile Development

Integrates Apple's FoundationModels framework into iOS applications for privacy-preserving, on-device text generation and structured data extraction.

This skill equips Claude with the specific implementation patterns required to utilize Apple's FoundationModels framework in iOS 26+ environments. It enables developers to build sophisticated AI features—such as text summarization, tool calling, and guided structured output—that run entirely on-device. By focusing on privacy and offline availability, the skill provides guidance on managing model sessions, handling device eligibility checks, and using the @Generable macro for type-safe, real-time snapshot streaming in SwiftUI.

主要功能

01Custom tool calling patterns for connecting on-device AI to local app logic.

02Robust model availability and device eligibility handling logic.

030 GitHub stars

04Real-time snapshot streaming for responsive, progressive UI updates.

05Guided generation using the @Generable macro for structured Swift data types.

06Implementation patterns for Apple's FoundationModels framework and LanguageModelSession.

使用场景

01Developing privacy-sensitive iOS apps that process user data locally without cloud dependencies.

02Building interactive SwiftUI components that stream structured generative content in real-time.

03Implementing AI-powered features like natural language form filling or data extraction that function offline.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry foundation-models-on-device

For use in Claude.ai and ChatGPT

主要功能

01Custom tool calling patterns for connecting on-device AI to local app logic.

02Robust model availability and device eligibility handling logic.

030 GitHub stars

04Real-time snapshot streaming for responsive, progressive UI updates.

05Guided generation using the @Generable macro for structured Swift data types.

06Implementation patterns for Apple's FoundationModels framework and LanguageModelSession.

使用场景

01Developing privacy-sensitive iOS apps that process user data locally without cloud dependencies.

02Building interactive SwiftUI components that stream structured generative content in real-time.

03Implementing AI-powered features like natural language form filling or data extraction that function offline.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry foundation-models-on-device

For use in Claude.ai and ChatGPT