skills.rest
scouzi1966
scouzi1966COMMUNITY
Explore all skills by this author →
Website𝕏@scouziGitHub

afm

Launch a local macOS ML stack with an OpenAI-compatible API.

324|17|Updated Aug 8, 2025
Search Tags:#macos#mlx#on-device#gateway#openai-compatible#vision-ocr#local-ml
One-click install
npx skills add https://github.com/scouzi1966/maclocal-api --skill afm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: afm
Source: https://github.com/scouzi1966/maclocal-api/tree/main/skills/afm
Command: npx skills add https://github.com/scouzi1966/maclocal-api --skill afm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

AFM provides a local, OpenAI-compatible API server and CLI to host Apple Foundation Models, MLX backends, and Vision OCR, enabling on-device inference, API routing, and secure local workflows without cloud keys.

Core Features & Use Cases

  • Local on-device inference for Foundation and MLX models on macOS.
  • OpenAI-compatible endpoints for /v1/chat/completions and /v1/models with gateway/proxy capabilities.
  • Vision OCR, tool-calling, and streaming responses for offline or secure environments.

Quick Start

Build and run AFM locally to start serving via the afm CLI and connect a client to the local API.

Frequently Asked Questions about afm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run local ML inference on macOS with an OpenAI-compatible API?▼

You can run local ML inference on macOS with an OpenAI-compatible API by launching a local server that hosts Apple Foundation and MLX backends, exposing /v1/chat/completions endpoints for on-device model routing without cloud keys.

Can I use MLX models for offline OCR and tool-calling in secure environments?▼

Yes, MLX models support offline Vision OCR and tool-calling in secure environments by routing requests through a local API gateway, enabling streaming responses and on-device inference workflows without external network dependencies.

What's the best way to build an on-device inference pipeline for macOS Foundation Models?▼

The best way to build an on-device inference pipeline for macOS Foundation Models is deploying a local API gateway with a vendor patch workflow, supporting streaming responses and tool-calling directly through OpenAI-compatible endpoints.

Does macOS local ML support OpenAI API endpoints like /v1/models and /v1/chat/completions?▼

Yes, macOS local ML supports OpenAI API endpoints including /v1/models and /v1/chat/completions, allowing developers to connect existing OpenAI-compatible clients directly to local Foundation and MLX model backends.

Why use a local model gateway instead of cloud APIs for on-device inference?▼

Use a local model gateway instead of cloud APIs for on-device inference to eliminate cloud key dependencies, ensure secure offline operation, and apply vendor patch workflows for deploying MLX and Foundation models in restricted environments.

Related Skills

Explore More Software Engineering Skills→
onymchatonymchat

apple-on-device-ai

Guide framework selection and model optimization for on-device AI on Apple Silicon.

Official
Advanced
copilot-ldcopilot-ld
1

libllm

Expose an HTTP LLM API client for OpenAI-compatible chat completions and embeddings.

Official
Advanced
DFly7DFly7

apple-on-device-ai

Select, convert, and optimize on-device AI models for Apple platforms.

Community
Advanced
rawvegrawveg
40

ollama

Run and interact with local AI models via the Ollama API.

Community
Advanced