Claude Code Complete Setup Guide: Terminal-Native AI for Network Security Automation

After years of using browser-based AI tools for network security documentation and automation scripts, I’ve found Claude Code to be a game-changer. Unlike the web interface, it runs natively in your terminal with direct file system access and can actually execute tasks on your local machine.

For network engineers dealing with firewall configs, automation scripts, and documentation, this setup transforms how you interact with AI. Instead of copying code snippets back and forth, Claude can read your actual config files, modify them, and even control browsers for testing.

What Makes Claude Code Different Link to heading

Claude Code is Anthropic’s terminal-native coding assistant. The key difference from browser interfaces is file system access and tool integration. When I’m working on Palo Alto automation scripts or n8n workflows, Claude can directly read my project files, understand the context, and make changes without manual copy-paste cycles.

Here’s the comparison that matters for our work:

Interface	File Access	Best Use Case
Claude.ai (browser)	None	Research, documentation
Claude Code CLI	Full local access	Building, automating, scripting
VS Code Extension	Limited	Quick inline questions

The real power comes from MCP (Model Context Protocol) servers - think of them as plugins that give Claude specific capabilities like browser control or GitHub integration.

Installation and Setup Link to heading

Prerequisites and Base Installation Link to heading

First, install Node.js LTS from nodejs.org. This is required for the Claude Code CLI:

# Verify installation
node --version   # Must be 18 or higher
npm --version

Install Claude Code globally:

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

# Verify and authenticate
claude --version
claude   # Opens browser login on first run

### Storage Optimization for Windows

To keep your C drive clean, redirect npm storage to D drive:

```bash
npm config set cache D:\npm-cache
npm config set prefix D:\npm-global

Add D:\npm-global\bin to Windows PATH:

Press Windows key → search ‘Environment Variables’
Edit system environment variables → Environment Variables
Under System Variables, find ‘Path’ → Edit → New
Add: D:\npm-global\bin

For Playwright browsers, create a system environment variable:

Variable name: PLAYWRIGHT_BROWSERS_PATH
Variable value: D:\playwright-browsers

mkdir D:\playwright-browsers
npx playwright install

Critical: Restart VS Code after setting environment variables.

MCP Server Configuration Link to heading

MCP servers are what make Claude Code truly powerful. I use four essential servers for network security work:

Installing All MCP Servers Globally Link to heading

The --scope user flag makes servers available across all projects:

# Playwright MCP - browser control for testing
claude mcp add --scope user playwright -- cmd /c npx -y @playwright/mcp@latest

# Filesystem MCP - read/write local files
claude mcp add --scope user filesystem -- cmd /c npx -y @modelcontextprotocol/server-filesystem C:\Users\robyk D:\

# Sequential Thinking MCP - complex task planning
claude mcp add --scope user sequential-thinking -- cmd /c npx -y @modelcontextprotocol/server-sequential-thinking

# GitHub MCP - repository management
claude mcp add --scope user github -e GITHUB_PERSONAL_ACCESS_TOKEN=YOUR_TOKEN -- cmd /c npx -y @modelcontextprotocol/server-github

Verify installation:

claude mcp list

Expected output:

playwright          ✓ Connected
sequential-thinking ✓ Connected  
github              ✓ Connected
filesystem          ✓ Connected

All MCP settings are stored in C:\Users\robyk\.claude.json. Never commit this file to GitHub - it contains your GitHub token.

Project Context with CLAUDE.md Link to heading

The CLAUDE.md file is crucial for consistent results. Place one in each project folder to give Claude context about your specific setup and requirements.

For a network automation project, here’s a sample CLAUDE.md:

## Project Overview
Network security automation scripts and Palo Alto firewall management

## Stack
- Python 3.11 with pan-os-python library
- Palo Alto firewall API integration
- n8n workflows for automation triggers
- MinIO for config backup storage

## Rules
- Never hardcode firewall credentials
- Always validate XML before committing changes
- Include rollback procedures in all scripts
- Test against dev firewall first

## File Structure
- /scripts/ - Python automation scripts  
- /configs/ - Firewall configuration templates
- /workflows/ - n8n workflow exports
- /docs/ - Network diagrams and documentation

Always launch Claude from your project directory:

cd D:\projects\network-automation
claude

Practical Use Cases for Network Security Link to heading

Browser Automation with Playwright MCP Link to heading

Playwright MCP excels at automating web-based network tools. I use it for:

Palo Alto GUI Automation:

Ask Claude: "Use Playwright to log into the firewall at 192.168.1.1, navigate to Policies > Security, and take a screenshot of the current rules"

Certificate Monitoring:

Ask Claude: "Check SSL certificate expiry dates for our critical websites and create a report"

Documentation Screenshots: For creating network documentation, Claude can automatically capture screenshots of network device interfaces, update diagrams, and maintain visual documentation consistency.

File System Integration Link to heading

With filesystem MCP, Claude can directly work with your configuration files:

Ask Claude: "Review all .xml files in the configs folder and identify any security policies that allow 'any' as source or destination"

Ask Claude: "Generate a Python script that backs up all firewall configs to MinIO, then update our backup inventory spreadsheet"

GitHub Integration for Change Management Link to heading

The GitHub MCP handles repository operations for configuration management:

Ask Claude: "Create a pull request for the firewall config changes, include the change summary and impact analysis in the PR description"

Session Management Best Practices Link to heading

Context Window Management Link to heading

Claude’s context window fills up during long sessions. Use these commands strategically:

/clear - Reset context but keep CLAUDE.md (use between different tasks)
/compact - Compress context when session gets heavy
/mcp - Show all connected servers and available tools

Planning Before Execution Link to heading

For complex automation tasks, always request a plan first:

"Do not write any code yet. Give me a detailed plan for creating an n8n workflow that monitors our firewall logs, identifies suspicious traffic patterns, and sends alerts to our SIEM."

Review the plan, provide feedback, then give the green light. This approach results in better code and helps you understand the implementation.

Troubleshooting Common Issues Link to heading

‘claude’ not recognized:

npm install -g @anthropic-ai/claude-code
# Verify PATH includes npm global bin directory

MCP Connection Failed: On Windows, ensure you use the cmd /c wrapper:

claude mcp add --scope user filesystem -- cmd /c npx -y @modelcontextprotocol/server-filesystem C:\Users\robyk D:\

Browsers downloading to C drive: Set the environment variable PLAYWRIGHT_BROWSERS_PATH=D:\playwright-browsers and restart VS Code completely.

Key Takeaways Link to heading

Claude Code transforms AI from a consultation tool into an active development partner. For network security engineers, the combination of file system access, browser automation, and GitHub integration creates powerful automation possibilities.

Start with simple tasks like config file analysis, then progress to complex automation workflows. The sequential thinking MCP is particularly valuable for planning multi-step network changes before implementation.

The Pro plan provides approximately 10-45 messages per 5-hour rolling window, shared across all Claude interfaces. Use /clear between unrelated tasks to optimize token usage.

Remember: always launch Claude from your project directory, use CLAUDE.md for consistent context, and plan complex tasks before execution. This setup has significantly improved my automation workflow efficiency and code quality.