This document outlines all required dependencies and setup steps needed to run the AI assistant both standalone (test mode) and inside MagicMirror².
Before running this project, ensure the following are installed on your system:
- Version: 18 or higher
- Check installed version:
node -v- Download: https://nodejs.org/
👉 Node 18+ is required for built-in fetch support.
- Comes with Node.js
- Verify installation:
npm -vThis project uses a local large language model via Ollama.
ollama pull llama3ollama run llama3👉 This must be running BEFORE starting the assistant.
- Fully supported ✅
- Built-in TTS works out of the box
- Supported
⚠️ - TTS depends on installed system voices
- May require additional speech engines (e.g.,
espeak)
This project uses the Node package:
npm install say- macOS → uses native
sayvoices - Windows → uses SAPI voices (may differ)
- Ensure system audio is enabled
Install all required dependencies:
npm installOr manually:
npm install say readline-syncTo run the assistant outside MagicMirror:
test.js
node test.js- AI will not respond
- Fix: run
ollama run llama3
- Check system volume
- Verify
sayworks manually:
say "Hello"Run:
npm installBefore running, ensure:
- Node.js 18+ installed
- npm installed
- Ollama installed
-
llama3model downloaded - Ollama running
- Dependencies installed (
npm install) - Audio working (for TTS)
Once all prerequisites are complete, you can:
- Run locally →
node test.js - Run in MagicMirror →
npm run start
This project is designed to:
- Run fully locally (no external APIs)
- Be cross-platform compatible
- Integrate seamlessly into MagicMirror²