Dev Environment & Foundations

1. What You'll Need

• Python 3.9 or later, check with python --version or python3 --version
• VS Code, download from code.visualstudio.com
• An Anthropic API key, obtained from console.anthropic.com
• ~10 minutes and a terminal

2. Install VS Code Extensions

Open VS Code and install these two extensions from the Extensions panel (Ctrl+Shift+X / Cmd+Shift+X):

• Python, publisher: ms-python. Enables Python language support, linting, and virtual environment detection.
• Jupyter, publisher: ms-toolsai. Lets you run .ipynb notebooks directly inside VS Code without launching a browser.

3. Create Your Project Folder

Pick a home for your lab files. All 10 modules will build on the same notebook, so one folder is all you need.

mkdir claude-architect-lab
cd claude-architect-lab

4. Set Up a Virtual Environment

A virtual environment keeps the SDK and its dependencies isolated from the rest of your system, important when you're working with multiple Python projects.

# Create the environment
python -m venv .venv

# Activate it
# macOS / Linux:
source .venv/bin/activate

# Windows (PowerShell):
.venv\Scripts\Activate.ps1

# Windows (Command Prompt):
.venv\Scripts\activate.bat

Your terminal prompt should now show (.venv) to confirm it's active.

5. Install Dependencies

pip install anthropic python-dotenv ipykernel

• anthropic, the official Python SDK for the Claude API
• python-dotenv, loads your .env file so the API key is never hardcoded
• ipykernel, registers your virtual environment as a Jupyter kernel so VS Code can find it

6. Get Your API Key

• Go to console.anthropic.com and sign in (or create a free account).
• Navigate to API Keys in the left sidebar.
• Click Create Key, give it a name like "architect-lab", and copy the key, it starts with sk-ant-.

You only see the full key once, so copy it immediately.

7. Store Your Key Safely

Create two files in your project folder:

ANTHROPIC_API_KEY=sk-ant-api03-YOUR-KEY-HERE

.env
.venv/
__pycache__/
*.pyc
.ipynb_checkpoints/

8. Create Your Lab Notebook

• Open your project folder in VS Code: File → Open Folder
• Create a new file named lab.ipynb (File → New File, then save with the .ipynb extension).
• VS Code will open it as a notebook. In the top-right corner, click Select Kernel.
• Choose Python Environments → select the .venv you just created. If it doesn't appear, run Python: Select Interpreter from the command palette first.

9. Hello Claude, Your First Cell

Add a new code cell and paste the following. Run it with Shift+Enter.

import anthropic
from dotenv import load_dotenv

load_dotenv()  # reads ANTHROPIC_API_KEY from your .env file

client = anthropic.Anthropic()

message = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=256,
    messages=[
        {"role": "user", "content": "In one sentence, what is the Claude API?"}
    ]
)

print(message.content[0].text)

If everything is configured correctly, you should see a one-sentence response from Claude. That's your environment confirmed.

10. Understanding the Response Object

Before moving to Module 2, it's worth knowing what the API returned. Add a second cell:

print("Stop reason:", message.stop_reason)
print("Input tokens:", message.usage.input_tokens)
print("Output tokens:", message.usage.output_tokens)
print("Model:", message.model)

• stop_reason, "end_turn" means Claude finished naturally. The exam tests this heavily.
• usage, input and output token counts used to calculate cost.
• model, confirms which model version actually ran your request.

API Signal to Remember: tool_use vs. end_turn

Agentic control flow starts with this one field:

• stop_reason == "tool_use" means the loop continues. Your application must execute the requested tool, append a matching tool_result, and call Claude again with the full updated history.
• stop_reason == "end_turn" means the loop terminates. Claude completed its turn without requesting another tool.

Do not check whether Claude said it is done. Use the structured stop_reason value.

Troubleshooting

Lab Exercise: Environment Verification & Response Anatomy

Objective: confirm your dev environment and master the identification of critical API signals like stop_reason and usage.

1. Stateless request: use the Messages API to send a "Hello Claude" prompt. Print the entire Message object.
2. Signal inspection: identify the stop_reason. Verify it is "end_turn". Explain why parsing response text for phrases like "I am finished" is a loop termination anti-pattern.
3. Token auditing: inspect the usage block. Identify input_tokens and output_tokens. Calculate the cost of the request, applying the 1.1x multiplier for Opus 4.6+ when inference_geo: "us" is used.
4. Signal simulation: manually modify a request to include a mock tool. Verify that the API returns stop_reason: "tool_use". Explain how this signal drives the next iteration of the loop.