Getting Started with Claude Code on Your VPS

Prerequisites and Requirements

Before starting, ensure you have the following:

Server Requirements

Resource	Minimum	Recommended
OS	Ubuntu 22.04+	Ubuntu 24.04 LTS
RAM	1 GB	2+ GB
Node.js	v18	v20 LTS

Required Accounts

RamNode VPS: Any Cloud VPS plan running Ubuntu or Debian
Anthropic API Key: Get one at console.anthropic.com

Installing Claude Code

Claude Code requires Node.js 18 or later. Here's the fastest path to a working installation:

Install Node.js 20 LTS

Install Node.js via NodeSource

# Install Node.js 20 LTS via NodeSource
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt-get install -y nodejs

# Verify installation
node --version  # Should show v20.x.x
npm --version   # Should show 10.x.x

Install Claude Code

Install globally via npm

npm install -g @anthropic-ai/claude-code

Authenticate

Run Claude Code for the first time to set up authentication:

First run authentication

claude

You'll be prompted to enter your Anthropic API key. Claude Code stores this securely in ~/.claude/ and you won't need to enter it again.

Verifying the Installation

Test that everything works correctly:

Test Claude Code

claude --version
claude "Say hello and tell me what OS I'm running on"

Claude Code will analyze your system and respond. You should see it identify your Linux distribution and kernel version.

Understanding How Claude Code Works

Claude Code operates differently from ChatGPT or the Claude web interface. Key differences:

It runs locally

Claude Code executes on your machine, not in a browser sandbox. When you ask it to create a file, it creates a real file. When you ask it to run a command, it runs a real command.

It has filesystem access

Claude Code can read your project files, analyze logs, and write new files—all with your permission.

It's conversational but stateful

Each session maintains context. You can iteratively refine scripts, debug issues step-by-step, and build on previous outputs.

It asks before acting

By default, Claude Code will show you what it plans to do and ask for confirmation before executing commands or writing files.

Your First Task: Server Hardening Script

Let's put Claude Code to work on something practical. Every new VPS should be hardened before production use. Instead of copying scripts from random blog posts, let's generate one tailored to your system.

Starting the Conversation

Start Claude Code session

claude

Once in the Claude Code session, describe what you need:

"I just spun up a fresh Ubuntu 24.04 VPS. Generate a comprehensive server hardening script that:

1. Creates a non-root sudo user
2. Configures SSH (disable root login, change port, key-only auth)
3. Sets up UFW with sensible defaults
4. Installs and configures fail2ban
5. Enables automatic security updates
6. Sets up basic system auditing

Make it idempotent so I can run it multiple times safely."

What Happens Next

Claude Code will:

Analyze your request
Generate a complete bash script
Ask if you want to save it to a file
Optionally offer to execute it (review carefully first)

Example Generated Script

server-hardening.sh

#!/bin/bash
# Server Hardening Script for Ubuntu 24.04
# Generated by Claude Code

set -euo pipefail

# Configuration - modify these values
NEW_USER="deploy"
SSH_PORT="2222"

# Colors for output
RED='\033[0;31m'
GREEN='\033[0;32m'
NC='\033[0m'

log_info() { echo -e "${GREEN}[INFO]${NC} $1"; }
log_warn() { echo -e "${RED}[WARN]${NC} $1"; }

# 1. Create non-root user with sudo access
create_sudo_user() {
    if id "$NEW_USER" &>/dev/null; then
        log_info "User $NEW_USER already exists, skipping..."
        return 0
    fi
    
    log_info "Creating user $NEW_USER..."
    useradd -m -s /bin/bash -G sudo "$NEW_USER"
    
    # Copy SSH keys from root if they exist
    if [[ -f /root/.ssh/authorized_keys ]]; then
        mkdir -p /home/$NEW_USER/.ssh
        cp /root/.ssh/authorized_keys /home/$NEW_USER/.ssh/
        chown -R $NEW_USER:$NEW_USER /home/$NEW_USER/.ssh
        chmod 700 /home/$NEW_USER/.ssh
        chmod 600 /home/$NEW_USER/.ssh/authorized_keys
        log_info "Copied SSH keys to $NEW_USER"
    fi
}

# 2. Harden SSH configuration
harden_ssh() {
    log_info "Hardening SSH configuration..."
    
    cat > /etc/ssh/sshd_config.d/hardening.conf << EOF
Port $SSH_PORT
PermitRootLogin no
PasswordAuthentication no
PubkeyAuthentication yes
AuthenticationMethods publickey
X11Forwarding no
MaxAuthTries 3
EOF

    if sshd -t; then
        systemctl reload sshd
        log_info "SSH hardened on port $SSH_PORT"
    else
        log_warn "SSH config validation failed, reverting..."
        rm /etc/ssh/sshd_config.d/hardening.conf
        return 1
    fi
}

# 3. Configure UFW firewall
setup_firewall() {
    log_info "Configuring UFW firewall..."
    apt-get install -y ufw
    ufw --force reset
    ufw default deny incoming
    ufw default allow outgoing
    ufw allow "$SSH_PORT/tcp" comment 'SSH'
    ufw allow 80/tcp comment 'HTTP'
    ufw allow 443/tcp comment 'HTTPS'
    ufw --force enable
}

# 4. Install and configure fail2ban
setup_fail2ban() {
    log_info "Setting up fail2ban..."
    apt-get install -y fail2ban
    
    cat > /etc/fail2ban/jail.local << EOF
[DEFAULT]
bantime = 1h
findtime = 10m
maxretry = 5

[sshd]
enabled = true
port = $SSH_PORT
maxretry = 3
EOF

    systemctl enable fail2ban
    systemctl restart fail2ban
}

# Main execution
main() {
    if [[ $EUID -ne 0 ]]; then
        echo "This script must be run as root"
        exit 1
    fi
    
    log_info "Starting server hardening..."
    apt-get update
    
    create_sudo_user
    harden_ssh
    setup_firewall
    setup_fail2ban
    
    log_info "Server hardening complete!"
}

main "$@"

Refining the Output

Claude Code shines in iterative refinement. Continue the conversation to add features:

SSH config validation before applying
A dry-run mode that shows what would change
Logging to /var/log/hardening.log

Cost Considerations

Claude Code uses the Anthropic API, which charges per token. Here's what to expect:

Task Type	Approximate Cost
Simple script generation	$0.01 - $0.05
Complex multi-file project	$0.10 - $0.50
Extended debugging session	$0.20 - $1.00

Monitor your usage at console.anthropic.com.

Tips for Minimizing Costs

Be specific upfront: Detailed initial prompts reduce back-and-forth
Use /clear between tasks: Resets context and avoids paying for irrelevant history
Set spending limits: Configure monthly limits in your Anthropic dashboard

Useful Commands Reference

Command	Description
`claude`	Start interactive session
`claude "prompt"`	One-shot command
`claude -p "prompt"`	Print output only (no file writes)
`/help`	Show available commands in session
`/clear`	Clear conversation context
`/cost`	Show token usage for current session
`Ctrl+C`	Cancel current operation

Quick Troubleshooting

"Command not found" after install

Fix PATH issue

# Ensure npm global bin is in PATH
export PATH="$PATH:$(npm config get prefix)/bin"
# Add to ~/.bashrc for persistence
echo 'export PATH="$PATH:$(npm config get prefix)/bin"' >> ~/.bashrc

API key issues

Re-authenticate

# Re-authenticate
rm -rf ~/.claude
claude

Permission denied on scripts

Make script executable

chmod +x your-script.sh

What's Next

You've installed Claude Code and generated your first infrastructure script. In Part 2, we'll tackle something more ambitious: Automating OpenStack Infrastructure with Claude Code—generating Terraform configurations, managing security groups, and deploying multi-tier applications through natural language.

Continue to Part 2

Series Overview Part 2: OpenStack Infrastructure