Esta página se muestra en inglés. Una traducción al español está en curso.
ConfigsJul 5, 2026·3 min de lectura

Leon — Open-Source Personal AI Assistant You Can Self-Host

A privacy-focused personal assistant that runs on your own server. Built with Node.js and Python, Leon understands natural language, executes skills (modules), and works offline without sending data to third-party services.

Listo para agents

Instalación lista para agent

Este activo puede instalarse después de elegir el runtime, revisar el plan y ejecutar el comando correspondiente.

Native · 98/100Política: permitir
Superficie agent
Cualquier agent MCP/CLI
Tipo
Skill
Instalación
Single
Confianza
Confianza: Established
Entrada
Leon
Comando de instalación directa
npx -y tokrepo@latest install 35caa829-786d-11f1-9bc6-00163e2b0d79 --target codex

Ejecutar después de confirmar el plan con dry-run.

Introduction

Leon is a self-hosted personal AI assistant designed to live on your own server. Unlike cloud-based assistants, Leon processes everything locally, keeping your data private. It uses a modular skill system to handle tasks like checking the weather, managing to-do lists, speed-testing your network, and more.

What Leon Does

  • Processes natural language commands through a local NLU engine
  • Executes modular skills (packages) for various tasks
  • Supports voice input via speech-to-text integration
  • Responds with text-to-speech for a conversational experience
  • Runs entirely on your own hardware with no cloud dependency

Architecture Overview

Leon consists of a Node.js server that handles HTTP/WebSocket communication and a Python skills engine that processes NLU and executes actions. The NLU pipeline classifies user intent and extracts entities, then dispatches to the appropriate skill module. Skills are isolated Python or Node scripts with their own dependencies. A TCP bridge connects the Node.js core with the Python NLU/skills layer.

Self-Hosting & Configuration

  • Requires Node.js 20+ and Python 3.11+ on the host machine
  • Use the CLI to scaffold and start the server in minutes
  • Configure STT (speech-to-text) and TTS providers in settings
  • Add or develop custom skill packages for your specific needs
  • Deploy with Docker for a containerized self-hosted setup

Key Features

  • Fully offline operation with no data leaving your server
  • Modular skill system for extensible functionality
  • Multi-language NLU with support for English and French
  • Voice interaction via integrated STT and TTS engines
  • Active development with a roadmap toward LLM-enhanced reasoning

Comparison with Similar Tools

  • Home Assistant — focuses on IoT/home automation; Leon targets personal task assistance
  • Mycroft — similar concept but project was discontinued; Leon is actively maintained
  • Alexa/Siri/Google — cloud-dependent and closed; Leon is private and self-hosted
  • OpenAI Assistants — API-dependent; Leon runs entirely locally
  • Jan — focuses on LLM chat; Leon emphasizes actionable skills and automation

FAQ

Q: Can Leon work completely offline? A: Yes, it can run fully offline using local NLU models and local TTS/STT engines.

Q: What programming language are skills written in? A: Skills are primarily written in Python, with Node.js support for certain integrations.

Q: Is there a mobile app? A: Leon is accessed via a web interface that works on any device with a browser, including mobile.

Q: Can I add custom skills? A: Yes, the skill SDK allows you to create custom packages with their own NLU training data and action logic.

Sources

Discusión

Inicia sesión para unirte a la discusión.
Aún no hay comentarios. Sé el primero en compartir tus ideas.

Activos relacionados