Setup Guide

Set up ELO-AGI

Install, authenticate, and start using ELO-AGI on your machine.

System Requirements

OS

Operating System

macOS 12+, Ubuntu 20.04+, Debian 10+, Windows 10+ (WSL)

Py

Python 3.9+

Check with python3 --version

HW

Hardware

2 GB+ RAM, no GPU required

Opt

Ollama (optional)

For CLI chat mode. Install from ollama.com

Installation

Choose your preferred installation method

The recommended way to install ELO-AGI. Automatically creates an isolated environment and adds neuro to your PATH.

macOS / Linux

terminal
curl -fsSL https://eloagi.com/install.sh | bash

Windows (WSL)

terminal
curl -fsSL https://eloagi.com/install.sh | bash

This creates a virtual environment at ~/.elo-agi/ and symlinks the neuro, neuro-demo, and neuro-bench commands to ~/.local/bin/. To upgrade, run the same command again.

Use pipx to install ELO-AGI in an isolated environment with CLI commands available globally.

Install pipx (if needed)

terminal
brew install pipx
pipx ensurepath

Install ELO-AGI

terminal
pipx install elo-agi

To upgrade: pipx upgrade elo-agi

Install with pip inside a virtual environment.

terminal
python3 -m venv .venv
source .venv/bin/activate
pip install elo-agi

To upgrade: pip install --upgrade elo-agi

Clone the repository and install in development mode with all extras.

terminal
git clone https://github.com/ezekaj/elo-agi.git
cd elo-agi
pip install -e ".[all]"

Extras: [dev] for testing tools, [llm] for LLM integration, [all] for everything.

After installation, verify it works:

terminal
elo --version

Quick Start

Run your first cognitive cycle

Python API

python
from neuro import CognitiveCore

core = CognitiveCore()
core.initialize()
core.perceive({"text": "Hello, world"})
core.think()
action = core.act()
print(action)

The CognitiveCore runs the active inference loop: perceive sensory input, think using free energy minimization across all 38 modules, and act based on the integrated cognitive state.

CLI

terminal
elo                      # Interactive session
elo info                 # System information
neuro-demo all           # Run all demos

Web REPL

Try in your browser

No installation required. Run Python with all 38 cognitive modules.

Open REPL

Run Benchmarks

Evaluate cognitive performance

terminal
neuro-bench -o report.html

Generates an HTML report with scores across 5 categories: Causal Reasoning, Compositional Abstraction, Continual Learning, Robustness, and Language Understanding. 627 tests total.

Uninstall

One-line install

terminal
rm -rf ~/.elo-agi ~/.local/bin/neuro ~/.local/bin/neuro-demo ~/.local/bin/neuro-bench

pipx

terminal
pipx uninstall elo-agi

pip

terminal
pip uninstall elo-agi

What's Next

Next Steps