Integrate Core ML models into iOS apps for on-device inference.
npx skills add https://github.com/Roy-wonji/claude-config --skill coreml-roy-wonji
Please help me install this Agent Skill. Skill: coreml Source: https://github.com/Roy-wonji/claude-config/tree/main/skills/swift-ios-skills/skills/coreml Command: npx skills add https://github.com/Roy-wonji/claude-config --skill coreml-roy-wonji
This Skill helps developers integrate and optimize Core ML models in iOS apps so they can load models efficiently, run reliable predictions, and tune compute performance without fragile boilerplate.
.mlmodelc/.mlpackage, support async loading and runtime compilation, and choose practical deployment strategies (bundle vs on-demand resources).MLFeatureProvider, including async and batch inference patterns.MLModelConfiguration compute units (CPU/GPU/Neural Engine), use MLComputePlan for per-operation device dispatch insight, and handle memory lifecycle safely.CVPixelBuffer, work with MLMultiArray, and use MLTensor (iOS 18+) for modern preprocessing pipelines.CoreMLRequest iOS 18+ or VNCoreMLRequest) to get correct image preprocessing and run image models end-to-end.Use the coreml skill to integrate Core ML inference into an iOS app by loading a compiled model asynchronously, configuring compute units, and running predictions from a CVPixelBuffer input.
High-intent search queries and answers about installing and using this skill.
To load and run Core ML models asynchronously, use Swift Core ML APIs to load compiled `.mlmodelc` or `.mlpackage` files and execute predictions via typed auto-generated classes or MLFeatureProvider. This supports async, batch, and stateful workflows for reliable on-device inference.
Configure Core ML compute units by setting options in MLModelConfiguration to target the CPU, GPU, or Neural Engine. You can use MLComputePlan for per-operation device dispatch profiling to optimize on-device machine learning performance and memory lifecycle.
Use Vision integration with VNCoreMLRequest or CoreMLRequest (iOS 18+) to handle image preprocessing end-to-end. This ensures correct data conversion from images to CVPixelBuffer before running predictions through your Core ML model.
Yes, you can use MLTensor (iOS 18+) for modern preprocessing pipelines and MLMultiArray for advanced tensor data handling. These Swift Core ML APIs support dynamic MLFeatureProvider inputs and stateful workflows for complex on-device inference.
On-device machine learning inference supports runtime compilation and practical deployment strategies including bundled models and on-demand resources. Async loading allows flexible integration of Core ML models into iOS applications without fragile boilerplate.
mintuzIntegrate on-device AI models in iOS apps using Foundation Models and MLX Swift.
ckorhonenOptimize CoreML models with quantization, palettization, and pruning.
dpearson2699Load Core ML models and configure compute units for iOS inference.
harshav167Load Core ML models and run predictions in Swift iOS apps.