JSPM

Lattice AI — local-first Digital Brain that keeps your knowledge durable across any AI model.

pi extension that warms/cools the Local AI Cat resident model over the Local API

Automatically translates JSON language files to other languages using Google Translate, AWS, Azure, DeepL, OpenAI, Hugging Face (cloud & local) or local OpenAI compatible server

Native MLX inference for Bun on Apple Silicon — local LLM server (OpenAI/Anthropic-compatible) and TypeScript library. No Python.

PipelineScore CLI — benchmark LLMs on your own hardware. Zero-config: `npx @pipelinescore/cli` auto-detects Ollama / LM Studio / llama.cpp / MLX and walks you through a run. Fully deterministic suite (no API key required), scored locally, published to the

Unified image generation + CLIP vision analysis. 11 providers (OpenAI, Replicate, Stability, FAL, Together, BFL, Google, Fireworks, Ollama, local, OKSLOP). Zero-shot classification, embeddings, and stock photo search.

Local-first MCP toolbelt: delegates summarise/classify/extract/transform tasks from any MCP client to a local oMLX inference server. Apple Silicon, MLX-accelerated, KV-cache persistent.

MLX Swift integration for React Native - InferrLM fork with enhanced features

Local MLX model launcher extension for Pi

React Native TurboModule wrapping the @dvai-bridge/ios + @dvai-bridge/android native SDKs behind a shared TS API.

Nitro module package

Your Local AI Harness

MCP Server for Claude Code, Cursor, Cline, Copilot, Github Copilot, Windsurf - Visual AI Agent Plan Execution, Approval Workflow, Plan Visualization, Agent Orchestration. See what your AI is thinking before it writes code. Works with Claude, GPT, Gemini,

🤖 Your MLX-powered coding companion for Mac - AI assistant with Mistral

DVAI-Bridge Capacitor plugin: MLX (Apple Silicon GPU/Neural Engine) LLM inference via mlx-swift-lm, with embedded HTTP server. iOS 17+ at link time; Apple-Silicon-only at runtime.

A simple, interactive CLI for fine-tuning small language models on Apple Silicon. No YAML configs, no complex flags - just an interactive CLI that guides you through the process. ⚒️

MCP Server for Claude Code, Cursor, Cline, Copilot, Github Copilot, Windsurf - Visual AI Agent Plan Execution, Approval Workflow, Plan Visualization, Agent Orchestration. See what your AI is thinking before it writes code. Works with Claude, GPT, Gemini,

Local UI-grounding specialist for hybrid AI agents. Screenshot + text target → strict JSON bbox. Qwen3-VL-2B LoRA, MLX 4-bit + GGUF + Ollama builds. Daemon, HTTP server, batch, confidence, eval. Drop-in for Claude Code, Codex, browser-use, Skyvern. Cuts G

Semantic image search, locally without Internet

OMLX provider extension for pi - Access self-hosted Qwen models through a local MLX/OpenAI-compatible server

Local vision (describe images), STT (speech-to-text), and TTS (text-to-speech). macOS uses MLX (Apple Silicon GPU). Windows/Linux use whisper, edge-tts, and transformers.

Run 4 AI models together on Apple Silicon. Get results that rival cloud AI. Pay nothing. Forever.

Context compaction plugin for OpenClaw - works with local models (MLX, llama.cpp) that don't report token limits

AI-powered session intelligence tool — turn screen recordings into structured work summaries

Compute embeddings of text/images with CLIP model

OpenClaw local TTS plugin powered by mlx-audio — zero API key, zero cloud dependency

LLM inference for Node.js powered by Apple MLX on Apple Silicon

Convert text to speech using Chatterbox TTS on Apple Silicon

Convert any document (PDF, DOC, DOCX) to clean Markdown for RAG

MCP Server for Cog - Semantic Code Intelligence

Local-first context engine for Claude: retrieves from your Gmail and injects relevant snippets into claude.ai requests, all on your Mac.

Ultra-fast TTS with ~90ms latency. Local text-to-speech for AI agents with 8 built-in voices.

Run 4 AI models together on Apple Silicon. Get results that rival cloud AI. Pay nothing. Forever.

The foundational Node.js bridge to Apple's MLX framework via C++. Native Apple Silicon acceleration with zero-dependency Node-API bindings.

Driver for the MLX90614 Infrared Temperature Sensor

Local AI scene partner & life assistant — sovereign AI for creators. Runs on Mac Mini with voice I/O, scene management, and proactive scheduling. No cloud. No content policies. Your hardware, your model, your rules.

Node.js bridge to Apple's MLX framework via Swift. Apple Silicon acceleration with zero-dependency Node-API bindings.

SEMO Solo — zero-setup personal AI agent (SQLite + Ollama/MLX/API). No Postgres, no Slack.

Native Large Language Model execution for Node.js on Apple Silicon. Fast, memory-efficient text generation, chat sessions, and tool-calling using MLX and Swift.

Semantic search for Obsidian notes using local embeddings (Ollama, LM Studio, MLX)

Pi coding agent provider for MLX local AI models on Apple Silicon

Store and search embeddings

Custom OpenCode MLX provider for Gemma 4 models

🐚 螺蛳壳里做道场 — Pure local LLM workflows that run on consumer Macs. No cloud, no API keys, just your machine.

MLX steering vectors — coming soon (Python/JS library)

Native Vision-Language Model support for Node.js. Process images and video with multimodal LLMs using MLX and Apple Silicon.

Run your favorite LLMs (Qwen, Gemma4, NemotronH, DharaAR, ...) at blazing-fast speeds on Apple Silicon, with multi-modal (image/audio/video) support, from JavaScript/TypeScript.