The Claude desktop app, and the optional API path.

Install the Claude desktop app and verify your tabs RECIPE

Before you begin

You have a working email address and a way to pay for the Anthropic Pro subscription (about $20/month, billed by Anthropic). If you are under 18, do this together with a parent. The split most students will use:

• Account email and login: the student's. This is the account you will use day to day to run the desktop app.
• Payment method: the parent's. A credit or debit card the parent owns and is comfortable being billed monthly against. Pro is month-to-month and the parent can cancel at any time from the Anthropic billing page.
• Practical setup: student creates the account and reaches the subscription screen, then hands the keyboard to the parent for the card entry. The student then continues with the install and tab-verification steps.

Steps

1. Go to https://claude.com/download✓ and download the desktop app for your operating system. Install it like any other app (drag to Applications on Mac; run the installer on Windows).
2. Launch the Claude desktop app. If you don't have an Anthropic account yet, click “Sign up,” create one with email + password, and verify the email. (Under-18 students: this is the student's account.)
3. The app will prompt you to choose a plan if you don't already have a paid subscription. Choose Pro — that's the tier this course is built around. (Under-18 students: parent at the keyboard for this step. The card stays with the parent.)
4. After signing in, you should see three tabs at the top of the app: Chat, Cowork, and Code. Click each tab once to confirm it loads:
   • Chat: opens to a familiar chat interface, similar to Claude.ai. No file access — just conversation.
   • Cowork: opens to a tasks view. There may be a brief onboarding flow the first time. The tab is asking you to give the agent a task it can work on independently.
   • Code: opens to a project picker. The tab will ask you to select a folder on your computer for the agent to work in. You don't need to pick one yet — Module 3 walks you through this in detail.
5. Smoke-test Cowork. Switch to the Cowork tab. Give it a tiny task: “Write me a one-paragraph note about what AI Architect Academy teaches.” It should respond with a short paragraph after a brief working pause. This confirms the Cowork tab is connected to the model and can produce output.
6. Smoke-test Code. Switch to the Code tab. Click the folder picker and select your ~/ai-architect-academy/✓ course working folder (the one you set up in Lesson 1.5). Type: “List the files in this folder.” The agent should list the contents, including my-first-loop.md. This confirms the Code tab can read your local files.

If all three tabs load and both smoke tests succeed, your desktop app is set up correctly and ready for the rest of the course.

Optional advanced — Set up an Anthropic API key and store it safely RECIPE

Before you begin

You already have an Anthropic account from Recipe 1 (the one you signed in to the desktop app with). The developer console is a separate billing relationship on that same account — per-token, capped — and is independent of your Pro subscription. Pro keeps working whether or not you set up an API key here. If you are under 18, do this together with a parent. The billing protocol mirrors Recipe 1:

• Account login: the student's. Use the same Anthropic account from Recipe 1.
• Payment method: the parent's. A credit or debit card the parent owns and is comfortable being billed monthly against. The parent stays the responsible adult for the bill, even though the student is the daily operator.
• Spend cap: agreed together. A number the parent is comfortable with as a worst-case month and the student is comfortable as a working ceiling. $5–10 is a reasonable starting cap; you can raise it later if real usage demands more.
• Practical setup: student signs in and reaches the billing screen, then hands the keyboard to the parent for the card entry. The parent sets the spend cap before stepping away. The student then continues with API-key generation in the next steps.

Steps

1. Go to https://console.anthropic.com✓ and sign in with your Anthropic account from Recipe 1. (If you don't have an account yet because you skipped Recipe 1, click “Sign up” and create one now — same email and password rules.)
2. Inside the developer console, find the Billing section. Add a payment method (credit or debit card). (Under-18 students: parent at the keyboard for this step. The card stays with the parent; the student does not memorize the number.)
3. Set your monthly spend limit. In Billing, set a hard cap — start with $5 or $10 for this course; you can raise it later if real usage demands more. This is the cap the provider enforces; once exceeded, API access stops until the next billing cycle or you raise it. (Under-18 students: agree on the number with the parent before setting it.)
4. Bookmark the usage dashboard. Inside the console, find the Usage (or Cost & Usage) section — it shows current month-to-date spend, broken down by day and by model. Open it now and bookmark it in your browser as Anthropic Usage. Plan to check it once a week during the course (and any time a pipeline runs longer than expected). The bill at the end of the month is too late; the dashboard is how you catch a runaway in days, not weeks.
5. Go to the API keys section. Click “Create key.” Give it a name like ai-architect-academy-course. Copy the key to your clipboard. The console will not show the full key again after you close this dialog — you have one shot. If you lose it, delete the key and create a new one.
6. Open VS Code. With your course working folder open, click “New File” in the sidebar (or use File → New File). Name it exactly .env — the leading dot is intentional; that's a Unix convention for “config file.”

   Heads up on Windows: Windows hides files starting with a dot by default. In the VS Code sidebar this is fine — it'll appear there. In File Explorer, enable View → Show → Hidden items if you want to see it outside VS Code.

7. In the new file, write one line with your key pasted in:

   ANTHROPIC_API_KEY=sk-ant-...(paste your full key here)

   Rules for .env files:

   • One KEY=value pair per line.
   • No spaces around the = sign.
   • No quotes around the value.
   • No blank lines between entries (some loaders are picky).
8. Save the file. Close it.
9. Verify the key works. Two ways to do this; pick one.

   Option A — verify by using it later (recommended). The simplest verification is to wait until Module 6 or 8 actually uses the key in a real script. If the script can read the key and make a request, the key works. If it can't, come back here and re-check the .env file for stray spaces or quotes around the value. This is the path most students should take — no extra test to run now.

   Option B — optional Python script test (click to expand). Belt-and-suspenders verification right now.

   No Python knowledge needed — just paste and run. Python should already be installed from Lesson 2.2.

   In your course working folder, create a new file called test_anthropic_key.py and paste:

   import os
   from pathlib import Path
   
   # Load the key from .env
   env_path = Path(".env")
   for line in env_path.read_text().splitlines():
       if line.strip() and not line.startswith("#") and "=" in line:
           key, value = line.split("=", 1)
           os.environ[key.strip()] = value.strip()
   
   # Make a test request
   import urllib.request
   import json
   
   req = urllib.request.Request(
       "https://api.anthropic.com/v1/messages",
       headers={
           "x-api-key": os.environ["ANTHROPIC_API_KEY"],
           "anthropic-version": "2023-06-01",
           "content-type": "application/json",
       },
       data=json.dumps({
           "model": "claude-haiku-4-5-20251001",
           "max_tokens": 64,
           "messages": [{"role": "user", "content": "Say hello in one short sentence."}],
       }).encode("utf-8"),
   )
   
   with urllib.request.urlopen(req) as response:
       result = json.loads(response.read())
       print(result["content"][0]["text"])

   Run it from your terminal: python3 test_anthropic_key.py✓ (Mac/Linux) or python test_anthropic_key.py✓ (Windows).

   You should see a one-sentence greeting print to the screen. If you see an authentication error, your key is wrong or not loaded; re-open the .env and confirm it has no stray spaces, no quotes, and the full key.

   What does this script do? It reads your .env file, sends a one-sentence hello to the API, and prints what comes back. You will not write code like this directly very often; an agent (or a custom script in Module 6 or 8) does this for you. The point of running it now is to confirm your key works.

Protect your .env

If you use git for any project — and Module 3 will introduce it in depth — create a file called .gitignore in the same folder and add one line: .env. This tells git to never track that file, so it can never end up on GitHub. Most editor templates already include .env in .gitignore by default. Confirm before you commit anything.

Optional reference — OpenAI follows the same pattern RECIPE

The OpenAI flow is functionally identical. platform.openai.com → Billing → API keys. The key is stored as OPENAI_API_KEY in the same .env file. This course does not assume OpenAI; it is listed here so you can see that the pattern is not vendor-specific.

If you set up an OpenAI key, your .env now has two lines:

ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...

Both follow the same four rules.

Optional advanced — Install the Claude Code CLI RECIPE

When you'd want the CLI

The Claude Code engine is already inside your desktop app's Code tab — same model, same agent, no separate install. The CLI version is for students who would rather work inside a terminal than a GUI: power users, students comfortable with the command line, or anyone who wants to script Claude Code into a larger automation flow. Most students will never need this. If you're not sure whether you want it, skip this recipe and stay with the desktop Code tab.

Steps

1. Confirm Node.js is installed. Claude Code requires Node.js (a runtime that runs many command-line tools, including this one). In your terminal, run node --version✓. If you see a version number ≥ 18, skip to step 2.

   If the command is not found, install Node from https://nodejs.org/en/download✓. Choose the LTS version — that stands for “Long-Term Support,” meaning it's the stable, maintenance-friendly release rather than the bleeding-edge one. Run the installer with default settings. After installing, close and re-open your terminal, then run node --version✓ again to confirm.

2. Install Claude Code via npm. In your terminal, run:

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

   What this command does: npm is Node's package installer. -g means “globally” — install the claude command so you can run it from any folder, not just the current one. @anthropic-ai/claude-code is the package's name. The whole command says: install Anthropic's Claude Code package globally.

3. In your terminal, cd into your course working folder.
4. Start Claude Code: type claude and press Enter.
5. On first launch, Claude Code will prompt you to choose an authentication method:
   • Subscription auth (recommended for most students): sign in with the Anthropic account from Recipe 1 through the browser window Claude Code opens. Your usage counts against the Pro subscription's shared limits, same as the desktop Code tab.
   • API key auth: paste the Anthropic API key from the optional advanced section earlier in this lesson when prompted. You'll be billed per token against the cap you set on your developer console.

   Both work. Subscription auth is the default for almost everyone. You can switch later by running claude /logout and signing back in.

6. Back in the terminal, Claude Code is now waiting for a prompt. Type: “List the files in this folder.” It should reply with the file listing.
7. Test a tiny directed edit: “Add a one-line comment at the top of my-first-loop.md that says 'Last updated via Claude Code on [today's date]'. Show me the diff first.” Confirm and apply.

Troubleshooting

Before you debug by yourself, surface the error to an AI. If any step above fails, copy the full error output and paste it into Claude.ai (or whichever chat AI you're using). Ask "What does this mean and how do I fix it on [your OS]?" The most common errors below already have known fixes; everything else is faster to diagnose by asking than by guessing.

• claude: command not found after npm install -g✓. Most common: your terminal's PATH wasn't refreshed after Node was installed. Close the terminal completely and open a new one. If still not found on Mac/Linux, run npm config get prefix✓ — the path printed should be on your $PATH. On Windows, this usually fixes itself on terminal restart.
• Authentication browser window doesn't open. Look in the terminal for a URL — Claude Code prints one as a fallback. Open it manually in any browser.
• “Node not found” after install. Restart your computer, not just your terminal. Some OS PATH updates need a full reboot to take effect.