Setting Up Claude & Claude Code

Go to claude.ai and sign up. Start with the Pro plan ($20/mo). Pro gives you extended thinking (Claude reasons through hard problems step by step), file uploads, project memory, and enough usage for daily building. If you find yourself hitting usage limits regularly, the Max plan ($100/mo) removes those caps. But start with Pro — you can always upgrade later.

Projects give Claude persistent context across conversations. Click Projects in the sidebar, then Create Project. Name it after your product. Upload key documents — your README, architecture notes, brand guidelines, database schema, anything Claude should always know about. Every conversation inside this project will have access to those files. Create one project per major product you're building.

Be specific about what you want. Instead of “build me a dashboard,” say “build a dashboard page that shows monthly revenue, active users, and churn rate, using Recharts for the graphs.” Share screenshots for UI work — Claude can read images and match designs. Use “think step by step” for complex problems where you want to see the reasoning. Ask Claude to explain its choices. And always iterate — the first response is a strong starting point, not the final answer. Push back, refine, and the output gets dramatically better.

Claude Code runs on Node.js, so you need it installed first. The cleanest way is through nvm (Node Version Manager), which lets you switch between Node versions easily. Run: curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash then open a new terminal window and run nvm install --lts. Verify it worked with node --version — you need 18 or higher.

Run npm install -g @anthropic-ai/claude-code in your terminal. This installs the claude command globally. You can now run it from any directory. If you get a permissions error, you may need to run with sudo or fix your npm global directory permissions (nvm avoids this issue).

Run claude in any directory. The first time, it will open your browser and prompt you to log in with your Anthropic account. This links the CLI to your subscription — your Pro or Max plan covers both the web app and Claude Code usage. Once authenticated, you won't need to log in again.

Navigate to your project directory in the terminal and run claude. It automatically indexes your codebase — it reads your file structure, understands your tech stack, and is ready to work. Ask it questions about your code: “What does the checkout flow look like?” Request changes: “Add a loading spinner to the submit button.” It edits files directly and shows you exactly what changed.

Use /plan to have Claude think through complex tasks before writing code — this prevents it from going down the wrong path on big features. Ask “What does this file do?” to understand existing code before modifying it. Give feedback like “That's close but change X” to iterate — Claude remembers the conversation context and refines its approach. It edits files in place, runs your tests, and can even commit changes to git.

Create a file called CLAUDE.md in your project root. This is Claude Code's instruction manual for your project — it reads it automatically at the start of every session. Include your tech stack, coding conventions, environment setup instructions, common commands (how to run dev server, run tests, deploy), and things to avoid. The more specific your CLAUDE.md, the less you have to repeat yourself. Think of it as onboarding docs for your AI pair programmer.

The most productive pattern: start each session with a clear goal. Use Claude.ai for research and planning first — hash out the approach, think through edge cases, decide on the architecture. Then switch to Claude Code for implementation — it works faster when the direction is clear. Review what it produces, give feedback, and iterate. Commit frequently so you can roll back if something goes wrong. Treat it like pair programming with a very fast colleague who never gets tired.