Claude Code: Your AI Partner

Module 2: Claude Code — Your AI Partner

You’ve got a vault. Now it’s time to give it an AI that can actually see inside it.

Claude Code is different from the browser-based AI tools you’ve probably used. It’s a command-line tool that runs in your terminal — and it can read, write, and navigate the files on your computer. When you run it inside your vault folder, it has access to every note you’ve written.

That changes what’s possible.

Instead of copying text into a chat window, you open a terminal, type claude, and start talking to something that already knows your projects, your notes, and your current context. The conversation is grounded in your actual situation, not a vacuum.

Prerequisites and Install

You need Node.js version 18 or higher. Check what you have with node --version. If it’s below 18 or missing, install it from nodejs.org — the LTS version is fine.

Then install Claude Code:

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

On first launch, run claude from any directory. It will prompt you to authenticate with your Anthropic account. Follow the steps — it opens a browser, you sign in, you come back to the terminal. Done.

Now navigate to your vault:

cd ~/Vault
claude

You’re in. The AI has access to everything in that folder.

CLAUDE.md: Your AI’s Briefing Document

The first thing you need to create is a file called CLAUDE.md in your vault root. This is the instruction file — every time Claude Code starts, it reads this file and uses it to calibrate every response.

Without it, Claude knows nothing about you. With it, Claude knows your projects, your preferences, how you like to work, and what you’re trying to accomplish.

Create CLAUDE.md in your vault root. Here’s a starting template:

## About Me

- I'm a [role] based in [city, country]
- I write in British English
- I use Obsidian for knowledge management and Claude Code for AI assistance
- My main areas of focus: [list 2-3]

## My Vault Structure

- 01 Inbox — quick captures, unprocessed
- 02 Projects — active work with deadlines
- 03 Areas — ongoing responsibilities
- 04 Resources — reference material
- 05 Archive — completed and retired

## How I Like to Work

- Keep responses concise unless I ask for depth
- When I ask "what should I focus on?", check my Projects folder first
- Use markdown formatting in responses
- Flag if you're uncertain rather than guessing

Adapt every line to your actual situation. The template is just scaffolding.

Your First Real Conversation

With your vault open and CLAUDE.md in place, try these prompts:

“What projects am I working on?” Claude will read your 02 Projects folder and summarise what it finds. If you’ve only got the newsletter note from Module 1, it’ll tell you that.

“Summarise what I’ve been thinking about this week.” Claude will look at recent notes and pull out themes. Early on there won’t be much — that’s fine. The point is seeing that it can.

“What should I focus on today?” This one is where the system starts showing its value. Claude will look at your projects, your recent notes, and your stated priorities and give you a grounded answer — not generic advice, but something based on what’s actually in your vault.

These conversations feel different from ChatGPT or Gemini. The AI isn’t making up a generic answer. It’s reading your actual files and responding to what’s there. That’s the foundation of the co-operating system.

Before moving on:

• Claude Code installed and authenticated
• CLAUDE.md created with your context
• At least one conversation completed about your vault contents

---

This module evolves as your system does.