Part 1 of 6

Introduction & Environment Setup

Understand Agent Zero's architecture and prepare a secure VPS foundation with SSH hardening, firewall configuration, and intrusion prevention.

10 minutes

Ubuntu 22.04/24.04

Agent Zero is an open-source framework for deploying autonomous AI agents that can write code, execute commands, search the web, and collaborate with other agents to accomplish complex tasks. Unlike cloud-based AI assistants, self-hosting Agent Zero gives you complete control over your data, unlimited usage without per-token fees, and the ability to run fully private AI workflows.

What is Agent Zero?

Agent Zero is a Python-based AI agent framework created by Jan Tomášek in March 2024. It differs from traditional chatbots in several important ways:

Autonomous Execution

Agent Zero doesn't just suggest code or commands; it executes them directly within an isolated Docker environment. Give it a task like "create a Python script that monitors disk usage and sends alerts," and it will write, test, and refine the code until it works.

Multi-Agent Cooperation

Complex tasks can be delegated to subordinate agents. The primary agent (Agent 0) breaks down problems and spawns specialized sub-agents, each maintaining clean context for their specific subtask.

Persistent Memory

Unlike stateless chat interfaces, Agent Zero remembers previous solutions, facts, and instructions. This memory system categorizes information and retrieves relevant context automatically.

Fully Customizable

The entire framework behavior is controlled by editable prompt files. There are no hard-coded limitations—you can modify how agents reason, communicate, and use tools by editing markdown files in the prompts/ directory.

Why Self-Host Agent Zero?

Running Agent Zero on your own VPS provides advantages that cloud AI services cannot match:

Privacy — Your prompts, data, and agent memory never leave your server. This matters for sensitive business automation, proprietary code generation, or any workflow involving confidential information.
Cost Control — Cloud LLM APIs charge per token. Heavy automation workflows can accumulate significant costs. With a self-hosted setup, your VPS cost is fixed, and you can optionally run local LLMs for zero API fees.
No Rate Limits — Your agents can work continuously without hitting usage caps or waiting for quota resets.
Customization — Modify any aspect of agent behavior. Add custom tools, integrate with internal systems, or create specialized agents for your specific use cases.
Reliability — Your AI infrastructure runs independently of third-party service availability.

Architecture Overview

Agent Zero uses Docker to create an isolated execution environment:

Architecture Diagram

┌─────────────────────────────────────────────────────────┐
│                      Your VPS                           │
│  ┌───────────────────────────────────────────────────┐  │
│  │              Docker Container                      │  │
│  │  ┌─────────────┐  ┌─────────────┐  ┌───────────┐  │  │
│  │  │ Agent Zero  │  │   Python    │  │  Bash     │  │  │
│  │  │  Framework  │◄─┤  Execution  │  │  Terminal │  │  │
│  │  └──────┬──────┘  └─────────────┘  └───────────┘  │  │
│  │         │                                          │  │
│  │         ▼                                          │  │
│  │  ┌─────────────┐  ┌─────────────┐  ┌───────────┐  │  │
│  │  │   Memory    │  │  Knowledge  │  │  SearXNG  │  │  │
│  │  │   Storage   │  │    Base     │  │  Search   │  │  │
│  │  └─────────────┘  └─────────────┘  └───────────┘  │  │
│  └───────────────────────────────────────────────────┘  │
│                          │                              │
│                          ▼                              │
│                 ┌─────────────────┐                     │
│                 │   LLM Provider  │                     │
│                 │  (Cloud/Local)  │                     │
│                 └─────────────────┘                     │
└─────────────────────────────────────────────────────────┘

The Docker isolation is critical—Agent Zero executes arbitrary code, so containing it prevents accidental damage to your host system. The framework communicates with LLM providers (OpenAI, Anthropic, Ollama, etc.) to power the AI reasoning, while all execution happens inside the container.

Hardware Requirements

Agent Zero's resource needs depend primarily on whether you run LLMs locally or use cloud APIs.

Cloud LLM Configuration (API-based)

If you plan to use OpenAI, Anthropic, Groq, or other cloud providers for inference, Agent Zero itself is lightweight:

Resource	Minimum	Recommended
RAM	2 GB	4 GB
vCPUs	1	2
Storage	20 GB	40 GB
Network	100 Mbps	1 Gbps

A RamNode 4GB Premium VPS ($24/mo) handles this configuration comfortably, with headroom for memory storage growth and concurrent agent sessions.

Local LLM Configuration (Ollama)

Running models locally requires significantly more resources. The requirements scale with model size:

Model Size	RAM Required	Example Models
7B parameters	8 GB	Mistral 7B, Llama 3 8B, Qwen2.5 7B
13B parameters	16 GB	Llama 2 13B, Qwen2.5 14B
30B+ parameters	32 GB+	Qwen2.5 32B, Mixtral 8x7B

For local LLM hosting, a RamNode 8GB or 16GB Premium VPS provides the memory needed for responsive inference. CPU-based inference is slower than GPU but entirely viable for async workflows.

Initial Server Setup

This guide uses Ubuntu 24.04 LTS, though Ubuntu 22.04 works equally well. The instructions assume you've provisioned a fresh RamNode VPS and have root SSH access.

Connect to Your Server

Terminal

ssh root@your-server-ip

Update System Packages

Terminal

apt update && apt upgrade -y

Create a Non-Root User

Running services as root is a security risk. Create a dedicated user:

Terminal

adduser agentzero
usermod -aG sudo agentzero

Set a strong password when prompted. Then switch to the new user:

Terminal

su - agentzero

SSH Key Authentication & Hardening

On your local machine, generate an SSH key if you don't have one:

Local Machine

ssh-keygen -t ed25519 -C "your-email@example.com"

Copy your public key to the server:

Local Machine

ssh-copy-id agentzero@your-server-ip

Harden SSH Configuration

Edit the SSH daemon configuration:

Terminal

sudo nano /etc/ssh/sshd_config

Apply these security settings:

/etc/ssh/sshd_config

Port 2222                       # Change from default 22
PermitRootLogin no              # Disable root login
PasswordAuthentication no       # Require key authentication
PubkeyAuthentication yes
MaxAuthTries 3
ClientAliveInterval 300
ClientAliveCountMax 2

Restart SSH to apply changes:

Terminal

sudo systemctl restart sshd

Important: Before closing your current session, open a new terminal and verify you can connect on the new port:

New Terminal

ssh -p 2222 agentzero@your-server-ip

Firewall Configuration

Ubuntu includes ufw (Uncomplicated Firewall). Enable it with rules for SSH and the Agent Zero web interface:

Terminal

sudo ufw default deny incoming
sudo ufw default allow outgoing
sudo ufw allow 2222/tcp comment 'SSH'
sudo ufw allow 80/tcp comment 'HTTP'
sudo ufw allow 443/tcp comment 'HTTPS'
sudo ufw enable

Verify the rules:

Terminal

sudo ufw status verbose

You should see:

Output

Status: active

To                         Action      From
--                         ------      ----
2222/tcp (SSH)             ALLOW IN    Anywhere
80/tcp (HTTP)              ALLOW IN    Anywhere
443/tcp (HTTPS)            ALLOW IN    Anywhere

Install Fail2ban

Fail2ban monitors logs and temporarily bans IPs that show malicious behavior:

Terminal

sudo apt install fail2ban -y

Create a local configuration:

Terminal

sudo nano /etc/fail2ban/jail.local

Add these settings:

/etc/fail2ban/jail.local

[DEFAULT]
bantime = 1h
findtime = 10m
maxretry = 5
banaction = ufw

[sshd]
enabled = true
port = 2222
filter = sshd
logpath = /var/log/auth.log
maxretry = 3
bantime = 24h

Start and enable fail2ban:

Terminal

sudo systemctl enable fail2ban
sudo systemctl start fail2ban

Check status:

Terminal

sudo fail2ban-client status sshd

Final Configuration

Set the Timezone

Terminal

sudo timedatectl set-timezone America/New_York

Replace America/New_York with your preferred timezone. List available options with timedatectl list-timezones.

Enable Automatic Security Updates

Terminal

sudo apt install unattended-upgrades -y
sudo dpkg-reconfigure -plow unattended-upgrades

Select "Yes" when prompted to enable automatic updates.

Verify Your Setup

Before proceeding to Part 2, confirm your environment is ready:

Terminal

# Check Ubuntu version
lsb_release -a

# Verify firewall is active
sudo ufw status

# Confirm fail2ban is running
sudo systemctl status fail2ban

# Check available disk space
df -h

# Check available memory
free -h

Your output should show:

Ubuntu 22.04 or 24.04 LTS
UFW active with your configured rules
Fail2ban active and monitoring SSH
Sufficient disk space (20+ GB free)
Your provisioned RAM available

What's Next?

Your VPS is now secured and ready for Agent Zero installation. In Part 2: Docker Installation & Core Deployment, we'll:

Install Docker and Docker Compose
Clone the Agent Zero repository
Launch the framework for the first time
Access the web interface

The isolated Docker environment ensures Agent Zero's autonomous code execution can't affect your host system, while your firewall and fail2ban configuration protect against external threats.

← Back to Series Part 2: Docker Deployment