Why This Matters#

The terminal is the primary way to work with Claude Code. Here you’ll assign tasks, receive responses, review proposed changes, and decide whether to accept them or not. In this lesson, you’ll learn how to confidently interact with Claude Code.

How to Assign Tasks#

Working with Claude Code is a regular conversation. You write in plain human language, and Claude understands and acts.

Starting Interactive Mode#

cd ~/my-project
claude

Now you’re in interactive mode — just type what you need:

explain the structure of this project

what technologies are used here?

where is the main entry point?

Entry point — the file where program execution begins. For example, index.js, main.py, or App.tsx.

Quick Commands (Without Interactive Mode)#

If you need to do one task without entering a conversation:

claude "fix the build error"

Or get an answer and exit immediately:

claude -p "explain what the main function does"

The -p flag (from “print”) means: respond and close the program.

How to Read Claude’s Responses#

Claude Code responds directly in the terminal. Responses come in several types:

1. Text Explanation#

Claude tells you something — for example, describes a project or explains code. Just read it.

2. Proposed File Change#

When Claude wants to edit a file, it shows a diff — a comparison of “before / after”:

- old_text = "Hello"
+ new_text = "Hello, World!"

Diff — from the word “difference.” Lines with - will be removed, lines with + will be added.

3. Proposed Command Execution#

Claude may suggest running a command — for example, installing a package or running tests. It shows the command and asks for permission.

How to Accept and Reject Changes#

This is the most important part: Claude Code does nothing without your permission (in standard mode).

When Claude proposes a change, you have a choice:

Example Conversation#

You: add a greeting function to main.py

Claude: I'll add a greet() function to main.py:

  def greet(name):
      return f"Hello, {name}!"

Apply changes? [y/n]

• Press y — the file changes
• Press n — nothing happens
• Type “name the function hello instead of greet” — Claude proposes a new version

💡 Tip: don’t be afraid to reject changes. You can always ask Claude to redo them. It doesn’t get offended 😄

Working Modes#

Claude Code operates in three modes:

1. Normal Mode (Default)#

Claude asks permission before every action. The safest option.

2. Plan Mode#

Claude first describes a plan — what it intends to do — and waits for your approval. Only then does it start working. Great for large tasks.

3. Auto-accept Mode#

Claude does everything without confirmation. Use only when you fully trust the task.

For beginners, normal mode is recommended — it’s the default.

Useful Commands in a Conversation#

Continuing a Conversation#

Closed Claude Code and want to pick up where you left off?

claude -c

The -c flag (continue) — continues the most recent conversation in the current folder.

Or choose from a list:

claude -r

The -r flag (resume) — shows a list of past conversations for you to choose from.

Practice#

Try these exercises:

1. Run claude in any folder
2. Ask: what's in this folder?
3. Request: create a file notes.md with the heading "My Notes"
4. Accept the changes
5. Request: add a list of three items to notes.md
6. Reject the changes and ask for a different version
7. Exit with the exit command

Lesson Summary#

✅ Tasks are given in plain language — just like chatting
✅ Claude shows changes and waits for your confirmation
✅ Lines with - are removed, with + are added (diff)
✅ y to accept, n to reject, or describe what to change
✅ claude -c continues the last conversation, claude -r lets you choose from a list

Next lesson: Working with Code →