Before you can build anything, you need to set up your tools. The primary tool is the Claude Code command-line interface, or CLI, which allows you to interact with Claude directly from your terminal. Since it's a Node.js package, you'll need Node.js (version 16 or higher) and npm installed on your system. Once those are ready, installation is a single command.

The -g flag installs the package globally, making the claude command available from any directory in your terminal. After the installation completes, the next step is to connect the CLI to your Anthropic account.

When you run your first claude command, the CLI will automatically open a browser window and guide you through an OAuth flow. This process securely grants the CLI permission to act on your behalf without you ever having to manually handle API keys. After you approve the request, Anthropic provides a security token that the CLI stores locally in a configuration file. This token is used for all subsequent requests, so you only need to authenticate once.

With the CLI installed and authenticated, you can now set up a dedicated project workspace. This is a crucial step because Claude Code maintains context based on the directory it's run from. Create a new folder for your client data gathering system and navigate into it.

Inside your new project directory, initialize it for Claude Code. This command prepares the workspace and, most importantly, creates a special file.

This creates a file named CLAUDE.md. This file is the key to maintaining context between sessions. You can fill it with information about your project's goals, tech stack, coding conventions, or file structure. Every time you start Claude Code in this directory, it will read this file first to get its bearings. A well-maintained CLAUDE.md file drastically improves Claude's ability to provide relevant and accurate assistance.

Claude Code operates with the same permissions as your user account. When you ask it to create a file, delete a directory, or run a script, it uses the underlying permissions of your terminal session. Always be mindful of the commands you authorize it to run, just as you would when working in the terminal yourself.

To verify that everything is working, you can run a simple command to check the version or list the files in your current directory from Claude's perspective.

If these commands execute successfully, your local environment is configured and ready. You now have a direct line to Claude within your project, with a dedicated context file to guide its work.

Before moving on, let's review the key terms from this setup process.

Now, check your understanding of the setup process.

With your environment established, you're ready to start defining the structure of your data pipeline.