Your first conversation

Lesson outcome

You will have had at least one real conversation with Claude Code and will understand how the interface works — how to send messages, how Claude Code uses tools, and how the permission system keeps you in control.

Why this matters in an agency

If you have used ChatGPT or claude.ai before, Claude Code will feel familiar but different. It is not a chat window in a browser. It runs in your terminal, in a specific directory, and it can take action on your files and system. Understanding the interface and the permission model matters because Claude Code is not a toy — it is an operating tool. The permission system exists so that you stay in control of what it does on your machine.

Inputs, tools, and prerequisites

Claude Code installed and authenticated from the previous lesson. Your terminal open. A working directory — any folder on your computer. Your Obsidian vault folder is a good choice.

Step-by-step walkthrough

Navigate to a working directory

Before starting Claude Code, navigate to a folder where you want to work. In your terminal:

On Mac:

```
cd ~/Documents/AgencyVault
```

On Windows PowerShell:

```
cd C:\Users\yourname\Documents\AgencyVault
```

Replace the path with the actual path to your Obsidian vault. The directory you are in when you start Claude Code is the directory it operates in. It can read files there, create new files there, and explore subfolders from that starting point.

Start Claude Code

Type claude and press Enter. Claude Code starts up and shows you a prompt. You can now type messages.

Ask something simple

Start with a question that does not require any file operations:

```
What day of the week is it today?
```

Claude Code will answer. Notice how the interaction works — you type your message, press Enter (or the keyboard shortcut shown on screen), and Claude responds. This is the basic back-and-forth.

Ask Claude Code to look around

Now ask something that involves the file system:

```
What files and folders exist in this directory?
```

Something important happens here. Claude Code wants to use a tool — specifically, it wants to list the files in the current directory. Before it does that, it shows you a prompt asking for your permission. It displays the tool it wants to use and the parameters, then waits for your approval.

This is the permission system. Claude Code never takes action on your system without showing you what it plans to do first. You can:

• Allow the action (press Enter or type "y")
• Deny the action (type "n")
• Allow all similar actions for the rest of the session

For now, allow the action. Claude Code lists the files and then tells you what it found. If you are in your Obsidian vault, it should mention the test note and the Stack Versions note.

Understand tool use

Claude Code is not just a text generator. It has access to tools — specific actions it can perform on your system. The most common tools are:

• Read: Read the contents of a file
• Write: Create a new file or overwrite an existing one
• Edit: Make a targeted change to part of a file
• Bash: Run a terminal command
• Glob: Search for files by name or pattern
• Grep: Search for text inside files

When Claude Code wants to use a tool, it shows you the tool name, the parameters (like which file it wants to read), and waits for your approval. You decide whether to allow or deny each action. This is the control loop that makes Claude Code safe to use — it proposes, you approve, it executes.

Have a real conversation

Ask something more substantial:

```
Read the Stack Versions note in this vault and tell me what is in it.
```

Claude Code will use the Read tool to open the file and then describe its contents. Allow the tool use when prompted.

Try a few more questions. Ask Claude Code to explain something about the folder structure, or ask it a general question about your industry. Get comfortable with the rhythm: you ask, Claude proposes an action or responds directly, you approve if needed, and the conversation continues.

Exit Claude Code

When you are done, type /exit or press Ctrl+C to leave. Claude Code closes and you are back in your normal terminal prompt.

Failure modes and verification checks

The main failure mode is confusion about the permission prompts. If Claude Code asks to do something you do not understand, deny it and ask Claude Code to explain what it wants to do and why. It will explain in plain language. The other common issue is starting Claude Code in the wrong directory — if it cannot find your files, check which directory you are in.

Verification: you had at least one conversation where Claude Code used a tool (like reading a file) and you approved it. You understand the propose-approve-execute flow.

Implementation checklist

• Navigate to your vault folder in the terminal.
• Start Claude Code with the claude command.
• Ask a simple question that does not use tools.
• Ask a question that triggers tool use and observe the permission prompt.
• Allow the tool use and review the result.
• Exit Claude Code.

Immediate next action

Move to the next lesson. You are about to create, read, and modify files with Claude Code — the fundamental operations behind everything else in this course.

Exercise

Start Claude Code in your vault folder and ask it: "What would be a good folder structure for an agency vault?" Read the response but do not act on it yet — you will organize the vault properly in Module 10 with Claude Code's help. The point of this exercise is to practice having a conversation about your business. Notice whether the response is generic or specific to your situation. Since Claude Code has no context about your agency yet (no CLAUDE.md, no existing notes with business details), the response will be generic. That changes as you build the system through this course.