Documentation

Docs

Everything you need to understand, build, and run JARVIS OS. This documentation covers the build pipeline, system architecture, and getting started instructions.

Quick Start

Option A: Download ISO

The fastest way to try JARVIS OS. Download the pre-built ISO, flash it, and boot.

# Flash to USB
sudo dd if=jarvisos-*.iso of=/dev/sdX bs=4M status=progress

# Or boot in QEMU
qemu-system-x86_64 -m 4G -smp 4 -cdrom jarvisos-*.iso

Download page →

Option B: Build from Source

Clone the repo and run the full seven-script build pipeline on an Arch Linux host.

git clone https://github.com/jarvisos/jarvisos.git
cd jarvisos

# Run each build stage sequentially
sudo ./01-base-system.sh
sudo ./02-user-setup.sh
# ... through 07

View source ↗

Build System

The Seven-Script Pipeline

JARVIS OS is built through a sequence of seven modular scripts. Each script handles one stage of the build and can be audited independently.

Base System

Bootstrap the base Arch Linux system with essential packages and configuration.

User Setup

Create user accounts, set permissions, and configure sudo policies.

Desktop Environment

Install and configure KDE Plasma 6 with Wayland session defaults.

AI Stack

Install Ollama, configure local LLM inference, and set up model defaults.

Jarvis Core

Deploy the jarvis-core daemon, CLI tools, and session management.

MCP Integration

Set up the MCP server registry, tool permissions, and orchestration layer.

ISO Packaging

Package everything into a bootable ISO with Calamares installer.

Usage

The `jarvis` CLI

Once booted into JARVIS OS, interact with the AI through the jarvis command. The daemon connects to the local Ollama instance and has access to MCP tools based on your configured permissions.

# Start a session
jarvis

# The AI can now:
# - Create, modify, and delete system tools via MCP
# - Execute commands with configured privilege levels
# - Maintain conversational context across interactions

See jarvis-core for full CLI documentation.

Requirements

System Requirements

Minimum

CPU x86_64 processor
RAM 4 GB
Disk 20 GB
GPU Not required (CPU inference)