Getting started

Install, first launch, opening a folder, and connecting a model with a LEVEL11 account, your own API keys or a local model.

Install

Get the macOS (Apple Silicon or Intel) or Windows x64 build from the download page. On macOS open the .dmg and drag LEVEL11 into Applications; on Windows run the installer. A Linux build is in preparation.

The builds shipping today are not yet notarized by Apple or code-signed for Windows, so the first launch shows a warning once.

macOS: if it says the developer cannot be verified, open System Settings → Privacy & Security and click Open Anyway at the bottom. Control-clicking the app in Finder and choosing Open works too.
Windows: in the SmartScreen dialog click More info → Run anyway.

The app checks for a new version every six hours and applies a downloaded update on the next launch. Help → Check for Updates runs the check right away.

First launch: the guide

The first launch opens on one question: Which describes you best? The guide changes with the answer.

I have never used a coding tool — sign in by e-mail, then LEVEL11 downloads and installs what the computer lacks (Node.js to run apps, Git to keep a history of changes). No admin password, no terminal commands. Then describe what you would like to build in a sentence or two: a project is created in LEVEL11 inside your home folder and the agent starts right away.
I have used one a little — after signing in, a short tour of four essentials (tools, modes, the Auto model, rules and MCP).
I have used Cursor, Claude Code or similar — pick how to connect models (account, your own API keys, a local model) and read only what is different about LEVEL11. A few days later the start screen asks once for feedback.

Settings → Experience changes the answer, and Show the first-run guide again replays it.

The start screen

LEVEL11 opens on a Start tab. At the top is the Open Tool grid: Editor, DocView, Terminal, Browser, Agents, MCP, Integrations and Settings. Below it are Open Folder, Create a new project, Clone Git Repository, New File and your recent folders. The same tools are always available from the "+" next to the editor tabs.

All eight tools share one workspace, one session and one agent, so a conversation started in the editor continues in the terminal tool.

Open a folder

Agent sessions belong to a folder (the workspace) and are saved per folder. Until you open one, the agent panel says "Open a folder to start a session". Once a folder is open, the explorer, ripgrep search and terminals work from it, and the agent reads and writes files inside it.

Connect a model: three ways

Until a model is connected, the start screen shows a banner: "Connect a model to start." Pick one of these in Settings (⌘,).

1. A LEVEL11 account. Under Settings → LEVEL11 account, enter your e-mail and press Send sign-in code. A six-digit code arrives by mail; it is valid for ten minutes and allows five attempts. The first sign-in creates a Free account on the spot. There is no password. In this mode no provider key reaches your device: requests go through the LEVEL11 gateway within your plan's limits. See Models for what you can run and Account and billing for limits and prices.

2. Your own API keys. On the same screen choose Use my own API key and add an Anthropic (Claude), OpenAI (GPT) or Google (Gemini) key. The environment variables ANTHROPIC_API_KEY, OPENAI_API_KEY and GEMINI_API_KEY are recognised too. Keys are encrypted with the OS keychain (Electron safeStorage), stay on this device, and requests go straight to each provider. One key is enough for everything from model selection to agent runs; an account is optional.

3. A local model. Under Settings → Local model, connect an Ollama, llama.cpp or MLX server. No account, key or internet needed. The steps are in Local models.

The first conversation

Pick a mode (Chat, Builder or Auto-Run) and a model below the composer and send a question. ⌘P is quick open, ⇧⌘P the command palette. What the agent does from there is covered in Agent.