# Leon

> Open-source personal AI assistant built around tools, context, memory, and agentic execution that can run locally on your own server.

Leon is an open-source personal AI assistant created by Louis Grenard and released under the MIT license. It runs locally by default, uses dedicated tools, context and memory, and can handle tasks through deterministic skills or agent-style planning. The project is currently focused on a 2.0 Developer Preview on the develop branch.

## What It Is

Leon is a self-hosted personal assistant that can understand a goal, choose how to handle it, use tools, remember useful information, and recover when something goes wrong. It started in 2017 as a classic intent-classification assistant and is being rebuilt around tools, memory, context, and agent-style execution. The web app is available at localhost:5366 once running.

## How Execution Works

Leon offers three modes: smart mode chooses automatically, controlled mode follows deterministic native skills and actions, and agent mode plans step by step. Native skills follow a Skills -> Actions -> Tools -> Functions (-> Binaries) structure, while agent skills are backed by SKILL.md workflows. The README also describes progressively loaded computer use that can operate desktop and browser interfaces, verify visual outcomes, and run locally or through Satellite.

## Memory, Context, and Providers

Leon keeps layered memory covering durable preferences, day-to-day context, and recent discussion. It uses context about the user's environment to ground its answers. It supports both local and remote AI providers, and the website lists several cloud and offline text-to-speech and speech-to-text options such as Google Cloud, AWS, IBM Watson, CMU Flite, and Coqui STT.

## Setup Path

The 2.0 Developer Preview requires Node.js 24 or newer on Linux, macOS, or Windows. Setup involves cloning the repository, installing pnpm and dependencies, and running pnpm start, with pnpm run check to verify the setup. The README states the new documentation is not ready, and the current docs site mostly reflects the legacy architecture, which remains on the master branch.

## Current Status

The README notes that Leon is in a major transition period with 2.0 Developer Preview work on the develop branch, and contributors are being onboarded progressively. The legacy, more stable pre-agentic version is available on the master branch.

## Features
- Smart, controlled, and agent execution modes
- Native skills and SKILL.md-backed agent skills
- Tool use instead of plain text replies
- Computer use for desktop and browser interfaces
- Layered memory
- Environment context awareness
- Local and remote AI provider support
- Text-to-speech and speech-to-text support
- Modular skills, toolkits, bridges, and binaries
- Local web app and HTTP API

## Integrations
Google Cloud, AWS, IBM Watson, CMU Flite, Coqui STT, Node.js, Python

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, JETBRAINS_PLUGIN, CLI

## Pricing
Open Source

## Links
- Website: https://getleon.ai
- Documentation: https://docs.getleon.ai/
- Repository: https://github.com/leon-ai/leon
- EveryDev.ai: https://www.everydev.ai/tools/leon