Claude Code: 2026 Guide to Getting Started, Installing and Using It

Table of Contents

• What Claude Code is
• How to install Claude Code
• Claude Code plans and pricing
• Which model to use
• Extending Claude Code: skills, plugins and MCP
• Design and alternatives
• Claude Code compared with other terminal agents
• Where to start

Claude Code is Anthropic's coding agent: you describe what you want and it reads your project, edits the files, runs commands and checks the result. This guide is the starting point. It covers what it is, how to install it, which plan you need, which model to pick and how to extend it, and it sends you to the detailed article on each topic when needed.

It is written for people who code, but also for anyone automating their business who has never opened a terminal.

What Claude Code is

Claude Code is an agentic coding tool. It is not autocomplete: it understands the whole project and works across several files at once. In practice you ask for things like "write the tests for the payments module, run them and fix whatever fails", and it plans, changes the code and checks that it works.

It runs on several surfaces that share the same engine:

• Terminal (CLI): the full version.
• VS Code and Cursor: with in-editor diffs and @ mentions.
• JetBrains (IntelliJ, PyCharm, WebStorm…): with a diff viewer and selection context; requires the CLI to be installed.
• Desktop app: to review diffs visually and run several sessions at once.
• Web and mobile: at claude.ai/code, with nothing to install, handy for long tasks.

It also works with git: it creates branches, writes commit messages and opens pull requests.

How to install Claude Code

The native installer is the recommended route and updates itself in the background.

macOS, Linux and WSL:

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

Windows PowerShell:

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

Windows CMD:

curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

Other options: brew install --cask claude-code (Homebrew), winget install Anthropic.ClaudeCode (WinGet), the apt, dnf and apk repositories, or npm install -g @anthropic-ai/claude-code (requires Node.js 22 or later). Homebrew, WinGet and the Linux package managers do not update themselves by default.

When it finishes, open a new terminal and check the install with claude --version. Then go into your project and start it:

cd your-project
claude

The first time, it asks you to log in through the browser.

Requirements: macOS 13 or later, Windows 10 1809 or later (or Windows Server 2019+), Ubuntu 20.04+, Debian 10+ or Alpine 3.19+; 4 GB of RAM or more; an x64 or ARM64 processor; an internet connection, and being in a country Anthropic supports.

On Windows you can install it natively or inside WSL. If you go native, install Git for Windows: without it, Claude Code runs commands with PowerShell instead of Bash.

Instructions and requirements checked on October 8, 2026 in the official Claude Code documentation.

Claude Code plans and pricing

Claude Code has no free plan. To use it you need a Claude Pro, Max, Team or Enterprise account, or an Anthropic Console account, which charges for API usage. The free claude.ai plan does not include it. If your company already works with Amazon Bedrock, Google Cloud or Microsoft Foundry, you can also connect it there.

The subscription is the same one you use in the Claude chat: if you already pay for Pro or Max, Claude Code is included. Current prices for each plan are on the official Claude pricing page. Detailed prices and limits are in Claude Code pricing.

Which model to use

With a Pro, Max, Team, Enterprise or API account, Claude Code uses Opus 5.5 by default. You can change it at any time with /model inside the session, or at startup with claude --model sonnet. The most useful aliases are opus, sonnet, haiku and opusplan, which plans with Opus and executes with Sonnet.

The short rule: Sonnet for everyday work, Opus for long or delicate tasks. I compare the models with prices and use cases in Opus 5.5 vs Sonnet 5.5: which Claude model to use.

Extending Claude Code: skills, plugins and MCP

What sets Claude Code apart from a chat is that it adapts to the way you work. There are four pieces:

• CLAUDE.md: a file at the project root with your rules (style, architecture, libraries). Claude Code reads it at the start of every session. If your repository already has an AGENTS.md, it can read that one in place of a CLAUDE.md.
• Skills: instructions packaged in a SKILL.md that Claude loads when it needs them or that you run as a command.
• MCP: the open protocol for connecting Claude Code to external tools such as Google Drive, Jira, Slack or your analytics.
• Plugins: a package that bundles skills, subagents, hooks and MCP servers and installs in one go. You manage them with /plugin, and Anthropic's official marketplace is added automatically the first time you open Claude Code in the terminal.

Three examples with their own guide on this blog:

• Superpowers, so the agent plans, does TDD and reviews before delivering: how to install Superpowers in Claude Code.
• SearchFit SEO, for SEO audits and content from the terminal: SearchFit SEO in Claude Code.
• Amplitude via MCP, to ask your product analytics questions in plain language: Amplitude and Claude.

Design and alternatives

Claude Code writes code; for visual design Anthropic has a separate product, Claude Design, used from the browser or the desktop app. If you want something open or that runs on your machine, I compare two options in Claude Design alternatives: Open Design and MiniMax Design.

Claude Code compared with other terminal agents

Claude Code is not the only agent that lives in the terminal. The closest open alternative is OpenCode, which does not tie you to one model provider and has its own subscription, OpenCode Go. If you want to try it, start with the OpenCode installation guide and check OpenCode Go pricing and limits.

If you're torn between OpenCode and Claude Code, I compare price, models and limits in OpenCode vs Claude Code.

And beyond the tool, what matters most when coding with an agent is your judgment. I write about it in soft skills for coding with AI.

Where to start

1. Install Claude Code with the command for your system and check it with claude --version.
2. Open a real project and ask it to explain the code structure before changing anything.
3. Create a CLAUDE.md with three or four rules for your project.
4. Try a well-scoped task: a test, a bug, a small refactor.
5. Once you feel comfortable, install Superpowers and connect a tool through MCP.

What matters is starting with small tasks and reviewing every change. The agent is fast, but deciding what goes into your code is still your call.