Getting Started

Getting Started

Full cross-platform setup (Linux · macOS · Windows) is in INSTALL.md — or read the PDF version.

Prerequisites

• Python 3.10+ (all lessons)
• Node.js 18+ (Node.js ports — Lessons 1–2)
• .NET 8 SDK (C# ports — Lessons 1–2)
• Claude Code CLI on PATH (default AI provider — uses your existing login, no API key needed)

Clone and run

git clone git@github.com:nikolareljin/local-ai-lab.git
cd local-ai-lab

python -m venv venv && source venv/bin/activate
pip install -r requirements.txt

./run -l 1          # Lesson 1: RAG — launches web UI
./run -l 2          # Lesson 2: MCP server demo
./run -h            # full help

Switching AI providers

Set RAG_PROVIDER to use a different backend:

Provider	Value	Needs
Claude Code CLI (default)	claude	claude on PATH
Ollama (local, offline)	ollama	Ollama running + a pulled model
Gemini	gemini	GEMINI_API_KEY in .env
OpenAI	openai	OPENAI_API_KEY in .env

RAG_PROVIDER=ollama ./run -l 1 ask "How do I reset the device?"

Copy .env.example to .env to set keys persistently.