Skip to content

Getting Started

Requirements

  • Python 3.9+
  • OpenAI-compatible LLM backend with prompt-caching support
  • A model with strong tool/function calling (recommended: Gemma 4)

Installation

  1. Clone the repo: 
    git clone https://github.com/OrbFrontend/Orb.git
  2. Verify Python 3 is installed: python3 --version
  3. Enter Orb folder and start the app:
  4. Linux/Mac: ./run_unix.sh
  5. Windows: run_windows.bat

First Run

  1. Open the Endpoints sidepanel and configure your Writer and Agent LLM endpoints.
  2. The same model can serve both roles (suitable for local hosting).
  3. Two separate models give better results at higher token cost.

  4. Endpoints use a tree structure: each endpoint may have many models, each model has its own params and custom prompts.

  5. Create or import a character in the Characters tab.

  6. Click the character and send your first message.