/
← L1: Your First Claude API Call L3: Setup for UI, Backend & Knowledge Base →
Module 2: Setup Environment & Enterprise

L2: Claude Code — What It Is

L2: Claude Code — What It Is

Claude Code — What It Is and How to Set It Up

Up until this point, we have interacted with Claude through API calls where you build the surrounding architecture. However, Anthropic has released a powerful, first-party tool that packages Claude's agentic capabilities directly into your development workflow. This tool is called Claude Code.

1. What is Claude Code?

Claude Code is an agentic coding system that lives directly in your terminal, IDE (like VS Code or Cursor), or as a standalone desktop app.

Instead of copying and pasting code snippets between your editor and a web browser, Claude Code acts as an autonomous pairing partner. It doesn't just generate text; it is granted permission (with your oversight) to interact with your local environment.

What makes it "Agentic"?

Claude Code can:

  • Read your codebase: It actively searches directories to understand architecture and how modules connect without you manually uploading files.

  • Execute commands: It natively runs tools like the GitHub CLI, npm, or pytest.

  • Fix and Iterate: If it runs a test and the test fails, Claude Code can read the error logs, fix the bug, and re-run the test until it passes.

  • Manage Version Control: It can stage changes, write commit messages, create branches, and open pull requests directly via git.

2. The Architectural Philosophy

From an architectural standpoint, Claude Code represents a shift from "AI as a chat interface" to "AI as a continuous orchestrator."

  • The Fallback Protocol: Claude prioritizes structured API connectors. If it needs to do something complex, it prefers using an API. If none exists, it falls back to native shell commands.

  • Agent Safety & Permissions: By default, Claude Code operates cautiously. It will plan an approach, propose changes, and ask for your explicit approval before modifying files or running shell commands. You can approve individual actions or configure "Accept all" for trusted sessions.

  • Extensibility via MCP: Claude Code supports the Model Context Protocol (MCP). This means you can connect it to external data sources—like pulling Jira tickets or reading Google Drive design docs—directly from your terminal.

3. Prerequisites for Setup

Before installing Claude Code, you must ensure your environment meets the following requirements:

  • Operating System: macOS 13.0+, Windows 10/Server 2019+, or modern Linux distributions (Ubuntu, Debian, Alpine).

  • Hardware: At least 4GB of RAM and an x64 or ARM64 processor.

  • Account: You must have a paid Anthropic subscription (Pro, Max, Team, or Enterprise tier). Note: The free tier of Claude.ai cannot execute Claude Code commands.

  • Network: An active internet connection to communicate with Anthropic's hosted inference infrastructure.

  • Dependencies: If you are running natively on Windows, Git for Windows is required.

4. Installation Methods

Anthropic recommends the native installation method, which automatically manages background updates to keep you on the latest secure version.

Native Installation (Recommended):

  • macOS / Linux / WSL:

Bash curl -fsSL https://claude.ai/install.sh | bash

  • Windows (PowerShell):

PowerShell irm https://claude.ai/install.ps1 | iex

  • Windows (CMD):

DOS curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

Alternative Installation (Homebrew for macOS):

Bash

brew install --cask claude-code

5. Authentication and First Run

Once installed, Claude Code operates via the claude command in your terminal.

Step 1: Authenticate

Open your terminal in any project folder and type:

Bash

claude

On your very first run, it will intercept the command and prompt you to log in via your browser (/login). You will authenticate using your paid Anthropic account. Your credentials are securely stored locally.

Step 2: Start a Session

Once authenticated, running claude launches an interactive REPL (Read-Eval-Print Loop) session. You will see a welcome screen.

From here, you simply speak to it in plain language. For example:

  • > explain this project

  • > write unit tests for the auth module, run them, and fix any failures

  • > refactor database.py to use async functions

You can type /help at any time to see core commands, or press Tab for command completion. In Module 3, we will cover how to deeply configure Claude Code using CLAUDE.md to dictate project-specific rules and workflows.