What Is Claude Code? Install It and Use It Well (2026 Guide)

Claude Code is Anthropic's agentic coding tool. It reads your codebase, edits files, runs commands and works with your development tools, and it runs in your terminal, in VS Code and JetBrains, in a desktop app and in the browser. You describe what you want in plain language; Claude Code plans the approach, makes the changes across files and checks that they work.

This site was built with Claude Code, including the keyword research and the pages you are reading. Here is how to install it, what it is good at, and the setup that makes the difference between a toy and a teammate.

How to install Claude Code

The terminal version is the full-featured one. Anthropic's recommended install is the native installer, which also keeps itself updated in the background.

On macOS, Linux or WSL:

curl -fsSL https://claude.ai/install.sh | bash

On Windows PowerShell:

irm https://claude.ai/install.ps1 | iex

Prefer a package manager? On a Mac, brew install --cask claude-code installs the stable channel (Homebrew installs do not auto-update, so run brew upgrade claude-code now and then). On Windows, winget install Anthropic.ClaudeCode works the same way.

Open a new terminal and check it worked:

claude --version

Then start it inside any project:

cd your-project
claude

You log in on first run. If the ANTHROPIC_API_KEY environment variable is set, Claude Code asks you to approve that key instead. Most surfaces need a Claude subscription or an Anthropic Console account. If you don't have a key yet, my guide to getting a Claude API key takes two minutes.

Other ways to run Claude Code

VS Code and Cursor: search for "Claude Code" in the Extensions view. You get inline diffs, @-mentions and plan review in the editor.
JetBrains: install the Claude Code plugin from the JetBrains Marketplace. It uses the CLI, so install that first.
Desktop app: download Claude for macOS or Windows, sign in and open the Code tab. It includes Claude Code, so there is no separate CLI install, and it needs a paid plan.
Web: start at claude.ai/code with nothing installed. Good for long-running tasks and repos you don't have locally.

All of them run the same engine, so your CLAUDE.md, settings and MCP servers carry across.

What Claude Code is good at

The work I hand to Claude Code falls into a few buckets:

Features across many files. Describe the feature, let it plan, read the plan, then let it build. It traces how your code fits together instead of guessing.
Bugs with a stack trace. Paste the error. It follows the trail through the codebase to the root cause and proposes a fix.
The work you keep putting off. Tests for untested modules, lint fixes across a repo, dependency updates, release notes.
Git. It stages changes, writes commit messages, creates branches and opens pull requests.

You can also run it non-interactively with -p, which makes it composable with the rest of your shell:

git diff main --name-only | claude -p "review these changed files for security issues"

CLAUDE.md: the file that makes it yours

CLAUDE.md is a markdown file in your project root that Claude Code reads at the start of every session. Put the things you would tell a new teammate on day one in it:

The commands to build, test and run the project
Conventions: naming, folder structure, preferred libraries
What not to touch, and why
How you want commits and pull requests written

Keep it short and specific. "Use Prisma for all database access; never write raw SQL in route handlers" is useful. "Write clean code" is not. If your repo already has an AGENTS.md for other coding agents, Claude Code can read that too.

Claude Code and MCP

The Model Context Protocol lets Claude Code use tools outside your repo: your issue tracker, your database, your design files, or your own product. Adding a server is one command:

claude mcp add --transport http notion https://mcp.notion.com/mcp

Inside a session, /mcp shows connected servers and handles OAuth logins. I cover scopes, .mcp.json and Claude Desktop in detail in Claude MCP: how to connect MCP servers to Claude, and if you want to expose your own app, start with what an MCP server is.

Skills, hooks and subagents

Three features turn Claude Code from an assistant into a repeatable workflow:

Skills package a workflow your team repeats, such as /review-pr or /deploy-staging, so everyone runs it the same way.
Hooks run shell commands before or after Claude Code actions: format after every edit, lint before every commit.
Subagents split a big task across several agents working in parallel, with a lead agent merging the results.

When you want that same agent loop inside your own application rather than your terminal, that is the Claude Agent SDK.

How I use Claude Code to ship

My rule is that Claude Code writes code and I own the decisions. In practice:

Plan first on anything non-trivial. I ask for the plan, push back on it, and only then let it edit.
Small, verifiable steps. One feature per session, with tests or a running preview to prove it works.
Real feedback loops. Type checks, linters and tests in the project so it can catch its own mistakes.
Review every diff. It is fast, not infallible. The diff view in the IDE or desktop app makes this quick.

That workflow is how one developer can ship products like ServeStack and Pair without a team behind every layer.

Frequently asked questions

Is Claude Code free? No. Most surfaces need a paid Claude subscription or an Anthropic Console account with API credits. The CLI, VS Code and JetBrains also support third-party providers such as Amazon Bedrock and Google Cloud.

Does Claude Code work on Windows? Yes, natively through PowerShell or CMD, or inside WSL. Git for Windows is recommended so it can use Bash.

What is the difference between Claude Code and the Claude API? The Claude API is the raw interface for sending messages to Claude from your code. Claude Code is a finished agent on top of it with file editing, command execution, permissions and memory built in.

Can Claude Code use my own tools? Yes, through MCP servers, which you add with claude mcp add.

Want Claude built into your own product, not just your editor? That is what my AI integration services are for.