Skip to main content
This guide walks you through everything you need to go from zero to a working AI coding session. By the end you will have Luminy installed, a provider configured, a project indexed, and a live conversation with your codebase — all in under five minutes.
1

Download Luminy

Head to luminy.tech and download the installer for your operating system. Luminy is available for macOS, Windows, and Linux as a fully bundled native app — you do not need Node.js, Rust, or any other runtime installed on your machine.Once installed, Luminy keeps itself up to date automatically. The built-in auto-updater checks for new releases in the background and notifies you when an update is ready, so you never need to revisit the download page.
2

Launch the app

Open Luminy after installation. On first launch you will be greeted by a welcome screen that invites you to configure your first AI provider and create your first project. The sidebar on the left is where your projects and sessions will live. The main panel is the chat interface.
Luminy’s greeting rotates based on the time of day — don’t be surprised if it says good morning, good afternoon, or good evening when you open it up.
3

Configure a provider

Luminy needs access to an AI model before it can do anything useful. You have two main options:Option A — Ollama (local, free, private)If you want to run models entirely on your machine without any API key or internet connection, install Ollama and pull a model. In Luminy, open Settings → Providers, select Ollama, and confirm the local endpoint (defaults to http://localhost:11434). That’s it.Option B — Cloud provider (Anthropic, OpenAI, Gemini, and more)Open Settings → Providers, choose your provider, and paste in your API key. Luminy immediately validates the key and populates a list of available models for you to choose from.For the full list of supported providers and per-provider setup instructions, see Providers.
If you are trying Luminy for the first time and want a great local experience, pull qwen2.5-coder:7b via Ollama. It is fast enough to run on most modern laptops, has strong coding ability, and keeps everything entirely on your machine.
4

Add your first project

Click the + button at the top of the sidebar and select New Project. A file-picker dialog will open — choose the root folder of the codebase you want to work with.Luminy will begin indexing your project using Tree-sitter, building a structural map of your code across all supported languages (Python, JavaScript, TypeScript, Rust, Go, Java, C++, and C#). Indexing typically finishes in a few seconds for most projects. You will see a progress indicator in the sidebar while it runs.Once indexing is complete, your project appears in the sidebar and is ready for the AI to explore.
5

Start chatting

Select your new project in the sidebar and type your first message in the chat input. A good starting point is something open-ended like:
“Explain the overall architecture of this project.”
Luminy’s agentic loop will kick in: the AI will read entry-point files, trace imports, inspect directory structure, and synthesize a grounded explanation — all without you having to point it at anything. From there, you can ask it to make changes, fix bugs, write tests, or anything else your workflow demands.To learn about agent modes and how to get the most out of the agentic loop, see Understanding Agents.