applefoundationmodels
- On-Device Foundation Model: Apple''s Foundation Models framework
npx skills add majiayu000/claude-skill-registry --skill applefoundationmodels --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Uses Apple's on-device Foundation Models framework to perform natural language understanding, generate structured outputs, and support tool-assisted tasks within apps. It operates offline on-device and can run a ~3 billion-parameter model when Apple Intelligence is enabled on supported devices.
How it works
- Interactions occur through a stateful LanguageModelSession (or SystemLanguageModel) that can accept optional developer-provided Instructions to set the model's role or style. Sessions maintain a Transcript and avoid concurrent requests.
- Offers NLU capabilities such as classification, tagging, summarization, and entity extraction. Includes a Content Tagging adapter accessible via SystemLanguageModel(useCase: .contentTagging).
- Supports Guided Generation via the @Generable and related tooling to ensure outputs match a defined structure, with the framework handling JSON-like output through its generation process.
- Tool Invocation is supported by defining tools that conform to the Tool protocol; tools have a unique name and a descriptive explanation, with arguments defined as Generable types. Tools are attached to the session and the model may call them autonomously.
- Streaming responses are available via session.streamResponse, delivering snapshot objects that gradually fill the @Generable output type fields.
- Safety and guardrails are built-in; errors such as GenerationError.guardrailViolation are handled, and the model supports multiple locales (15+).
When to use it
Use when you need on-device NLP, offline operation, privacy-preserving processing, and structured outputs within Apple-native apps. Initialize a session only if the device supports Foundation Models and Apple Intelligence is enabled; consider using the contentTagging adapter for topic tagging and entity detection if appropriate to your domain. If real-time or incremental updates are desired, enable streaming responses.
What it can touch
- Tools: Defined by the app and attached at session initialization. Tools require a call(arguments:) method and a Generable argument type. The model will decide when to call a tool based on prompts and tool descriptions.
- Outputs: Generated as JSON-like structures that map to the @Generable output schema.
Caveats
- Requires iOS 26.0+, iPadOS 26.0+, macOS 26.0+, macOS Catalyst 26.0+, or visionOS 26.0+ with Apple Intelligence enabled. Availability must be checked via SystemLanguageModel.availability.
- On-device model is a ~3 billion-parameter model; performance and capabilities depend on device and OS support.
- Handling of sensitive content relies on built-in guardrails; errors like GenerationError.guardrailViolation may be raised.
- Results depend on correct prompt design and output schemas; ensure shapes defined by @Generable structs align with your app’s data models.
--- name: "AppleFoundationModels" description: "Use Apple's on-device **Foundation Models** framework (iOS/iPadOS/macOS 26.0+) for natural language understanding, structured data generation, and tool-assisted tasks in apps." version: "1.0" dependencies: - iOS 26.0 or later (Apple Intelligence enabled) - iPadOS 26.0 or later - macOS 26.0 or later - Mac Catalyst 26.0 or later - visionOS 26.0 or later --- # Instructions - **On-Device Foundation Model:** Apple's Foundation Models framework provides access to a \~3 billion-parameter large language model running entirely on-device[\[1\]](https://www.apple.com/newsroom/2025/09/apples-foundation-models-framework-unlocks-new-intelligent-app-experiences/#:~:text=The%20Foundation%20Models%20framework%20is,when%20Apple%20Intelligence%20is%20enabled). This model powers **Apple Intelligence** features and runs offline at no extra cost, ensuring user data stays private[\[2\]](https://www.apple.com/newsroom/2025/09/apples-foundation-models-framework-unlocks-new-intelligent-app-experiences/#:~:text=With%20the%20release%20of%20iOS,and%20help%20users%20in%20new). Always check that the device supports Apple Intelligence and that the model is available
What does the applefoundationmodels skill do?
- On-Device Foundation Model: Apple''s Foundation Models framework
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill applefoundationmodels --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.