Repo Agents Limited info

leon-ai/leon

Open-source personal AI assistant built around tools, context, memory and agentic execution; the 2.0 Developer Preview on the develop branch is the current focus.

  • 17.6k GitHub stars
  • TypeScript
  • ⚖️ MIT
  • 🎯 Advanced
leon-ai/leon preview image

What it is

Leon is an open-source personal AI assistant built around tools, context, memory, and agentic execution. It can run locally, use dedicated tools instead of free-form guessing, and work in smart, controlled, or agent modes. The repository's develop branch holds the 2.0 Developer Preview, while the master branch holds the older, more stable pre-agentic version.

Who it's for

  • Developers who want to explore or contribute to Leon's new 2.0 core on the develop branch
  • Users who want a privacy-aware assistant that can use local models and local context
  • Builders interested in extending an assistant through skills, toolkits, bridges, and binaries

Requirements

Requirements

  • Node.js >= 24.0.0 (Volta is recommended for managing Node.js)
  • Linux, macOS, or Windows
  • pnpm (installed via npm in the setup steps)

Setup

  1. Clone the repository and install dependencies

    Clone the repo, go to the project root, install pnpm globally, then install dependencies.

    sh
    # Clone the repository
    git clone https://github.com/leon-ai/leon.git
    
    # Go to the project root
    cd leon
    
    # Install pnpm
    npm install --global pnpm@latest
    
    # Install dependencies
    pnpm install
  2. Run Leon

    Start Leon. By default it runs locally and the app is available on http://localhost:5366.

    sh
    # Run Leon
    pnpm start
  3. Check your setup

    Verify that the setup went well.

    sh
    # Check the setup went well
    pnpm run check

Pros & cons

Pros

  • Pro:Can work with local models and local context, and also supports remote AI providers, so you can balance privacy, control, and capability
  • Pro:Offers three execution modes (smart, controlled, agent) so deterministic native skills and step-by-step agent planning are both available
  • Pro:Modular design with skills, toolkits, bridges, and binaries, plus layered memory and environment context
  • Pro:MIT-licensed and open source, so the architecture can be inspected and built on

Cons

  • Con:The new documentation for 2.0 is not ready; the docs site and older guides mostly reflect the legacy architecture
  • Con:The 2.0 Developer Preview is still in transition and moving, and the more stable pre-agentic version lives on the master branch
  • Con:Developed largely in spare time, so progress can be uneven and contributor onboarding is still gradual

Images