How to Use Claude Code in 2026: A Beginner's Guide from Install to Plans and Cloud Sessions

Claude Code is Anthropic's agentic coding tool that reads your project's code, edits files and runs commands from plain-language instructions. As of September 2026 it runs in the terminal, in IDEs such as VS Code and JetBrains, in the desktop app, in the browser and in the mobile app, and it is not part of the Free plan, so you need Pro ($20 a month) or higher, or an API account. Installing takes one command. Where beginners actually get stuck is not installation but which plan they need, how much they can use it, and how much work to hand over. This guide walks through that order using only Anthropic's official docs and announcements.

What is Claude Code?

Claude Code is Anthropic's AI coding agent that understands your whole codebase and works across multiple files and tools to build features, fix bugs and automate development tasks. Unlike pasting code into a chat window and copying answers back, Claude Code opens, edits and tests files directly inside the project folder on your computer. Your job is to describe what you want and review the changes Claude proposes or makes.

As of September 2026, the default Opus model in Claude Code is Opus 5.5, released on September 22, and Fable 5.1 is also available with a 1M-token context window. Anthropic says Opus 5.5 performs at the level of Fable 5.1 on most work and costs 40% less to run than Opus 5. To compare models on terminal tasks, see the [agentic terminal leaderboard](/leaderboard/agentic-terminal/).

How to install Claude Code in 5 steps

Installing Claude Code requires macOS 13.0 or later, Windows 10 1809 or later, or Ubuntu 20.04 or later, and takes a single official install command. You need 4 GB or more of RAM and an internet connection. Condensed from the official quickstart, the steps are:

1. Open a terminal. On macOS, Linux or WSL run curl -fsSL https://claude.ai/install.sh | bash; in Windows PowerShell run irm https://claude.ai/install.ps1 | iex. Homebrew users can run brew install --cask claude-code, and WinGet users can run winget install Anthropic.ClaudeCode.
2. Run claude --version and check that it prints a version number followed by (Claude Code).
3. Move into your project folder and type claude to start your first session.
4. On first launch you are prompted to log in. Sign in through the browser with a Pro, Max, Team or Enterprise subscription or a Claude Console (API) account.
5. Type /init inside the session. Claude analyzes your codebase and drafts a CLAUDE.md file with build commands, test instructions and project conventions.

Native installs update automatically in the background, but Homebrew and WinGet installs do not, so you need to run the upgrade command yourself. On Windows, installing Git for Windows lets Claude Code use its Bash tool. If the terminal feels unfamiliar, you can start with the desktop app instead.

Claude Code pricing: it is not included in the Free plan

Claude Code is not included in the Claude Free plan as of September 2026 and is included in every paid plan, from Pro ($20 a month) through Max, Team and Enterprise. According to Anthropic's official pricing page, Claude Code shares the same usage limits as the rest of your plan. Every hour you spend coding in the terminal draws from the same pool as your Claude chats.

Plan	Price (USD)	Claude Code	Usage per 5-hour session	Cloud sessions
Free	$0	No	Baseline	No
Pro	$20/mo	Yes	At least 5x Free	Yes
Max 5x	$100/mo	Yes	5x Pro	Yes
Max 20x	$200/mo	Yes	20x Pro	Yes
Team	From $20 to $25 per seat/mo	Yes	More than Pro	Yes
Console (API)	Prepaid credits, billed per token	Yes	Pay per use instead of limits	No

When you hit a limit, you can wait for the 5-hour session to reset, move to a higher plan, or turn on usage credits and keep going at standard API rates. If the ANTHROPIC_API_KEY environment variable is set, Claude Code bills that API key instead of your subscription, so check it if you mean to use your plan. Plan limits and resets are covered in detail in our [Claude pricing plans guide](/claude-pricing-plans-2026/).

Claude Code in the terminal, IDE, desktop and web compared, plus cloud sessions

Claude Code is the same agent wherever you run it, and what changes is where the code runs and whether your local configuration is available. The terminal, IDE and the desktop app in Local mode run on your machine, while the web (claude.ai/code), the Code tab in the mobile app and the desktop app in Cloud mode run on virtual machines managed by Anthropic.

Surface	Runs on	Best for
Terminal CLI	Your machine	Developers who want every feature
VS Code, JetBrains	Your machine	Reviewing changes inside the editor
Desktop app	Local or cloud	Beginners who prefer a graphical app
Web, mobile	Cloud VM	Handing over a GitHub repo with no install
Cloud sessions (`claude --cloud`)	Cloud VM	Long tasks that keep running with the laptop closed

Cloud sessions are available on Pro, Max and Team plans, and to Enterprise users with premium seats or Chat + Claude Code seats. You need to connect GitHub to start a session, and usage shares the same limits as the rest of Claude with no separate compute charge for the cloud VM. Anthropic's official developer account announced that, with cloud sessions leaving research preview, existing subscribers get a one-time credit of $100 on Pro and $250 on Max. The credit is separate from usage limits, applies only to cloud sessions, is limited to one per account and must be claimed by October 7.

Claude Code features: CLAUDE.md, skills, subagents and MCP

The core of Claude Code's extension layer is four features, CLAUDE.md, skills, subagents and MCP, which together cover most of what a beginner needs. The official docs also recommend starting with CLAUDE.md and adding the others only as specific needs come up.

• CLAUDE.md: a project instruction file that is loaded automatically at the start of every session. Put rules such as "Use pnpm, not npm" or "Run tests before committing" here. The docs suggest adding a rule once Claude gets the same convention wrong twice.
• Skills: markdown files that hold knowledge and workflows. You can call one directly with a command such as /deploy, or Claude loads it on its own when it is relevant.
• Subagents: helper agents that work in their own context and return only a summary. Handing them research that reads many files keeps the main conversation light.
• MCP: the Model Context Protocol connects Claude Code to external services, for example querying a database, posting to Slack or controlling a browser.

Beyond these, automation such as running a linter after every file edit is done with hooks, and reusing a bundle of these settings across repositories is done with plugins.

Starting vibe coding with Claude Code: what to know first

The most important habit for a beginner starting vibe coding with Claude Code is to get a plan before letting it write code. Since August 14, 2026, new sessions on Pro, Max and Team plans use auto mode as the default permission mode. In auto mode a classifier reviews Claude's actions instead of asking you each time, so a vague request turns into changes across many files that stay vague.

Pulling together the workflow the official docs recommend, ASAP sees three rules for beginners. First, start in plan mode, for example with claude --permission-mode plan, where Claude reads files and proposes an approach without editing source code. Read and adjust the plan, then let it execute, and you will undo far less. Second, commit to git after each unit of work. You can hand over more boldly when there is always a point to return to. Third, instead of "fix the tests", give the file name, the expected behavior and the error message. The official cloud session guide also recommends naming specific files and pasting error output.

The illusion vibe coding creates for beginners is that you no longer need to understand code. Claude Code writes the code, but a person still decides whether to accept the result, so the ability to read the changes is what sets the quality of what you ship. For which kinds of work pay off in practice, see [7 ways to use Claude Code at work](/claude-code-at-work/).

Usage limits are a beginner's real cost

The first wall Claude Code beginners hit is not a missing feature but the 5-hour session limit. Claude Code draws from the same usage pool as chat, and long conversations, large files, top models and parallel tasks all raise consumption. The official docs also note that running several cloud sessions at once uses up limits proportionately.

For beginners on Pro, then, cutting consumption comes before upgrading. Open a new session when a task is done, write rules you keep repeating into CLAUDE.md once, and hand file-heavy research to a subagent so the main conversation stays light. Simply not repeating the same explanations already lowers usage. Check remaining usage with /status in Claude Code or in Settings > Usage on the web. If you still hit the limit all through the workday after that, it is time to move to Max.

Sources: Claude Code overview (Claude Code Docs) · Quickstart (Claude Code Docs) · Advanced setup (Claude Code Docs) · Extend Claude Code (Claude Code Docs) · Use Claude Code in the cloud (Claude Code Docs) · What's new (Claude Code Docs) · Plans & Pricing (Anthropic) · Use Claude Code with your Pro or Max plan (Claude Help Center) · Introducing Claude Opus 5.5 (Anthropic) · Cloud sessions launch announcement (Anthropic @ClaudeDevs)