
Claude Code is Anthropic’s agentic coding tool for your terminal: it reads your codebase, plans a change, edits files, runs commands, and fixes its own mistakes.
It is not autocomplete. Autocomplete finishes the line you are typing. Claude Code takes a task in plain English, carries it through several files, and you review the result before anything is committed. Scrimba calls the explore, plan, code, commit loop the core workflow, and it is the habit to learn first.
What Claude Code is
Claude Code runs in your terminal and works directly on your project. You type a request, and it searches the files, makes the edits, runs the tests, and reports what changed.
The default model is Opus 5.5, with a one-million-token context window for large codebases. Switch models inside a session with /model, or start with claude --model sonnet for everyday work. Sonnet answers faster and costs less on API billing, so keep Opus for long or delicate tasks.
The agentic style is the difference: it acts on the project instead of only suggesting text. The AI coding assistant guide compares the main options, and the Claude vs ChatGPT comparison covers the chat models behind these tools.

How to install Claude Code and run your first project
The free plan does not include Claude Code, so you need a paid plan or a Console account first. On macOS, Linux, or WSL, run curl -fsSL https://claude.ai/install.sh | bash. On Windows PowerShell, run irm https://claude.ai/install.ps1 | iex. The npm route, npm install -g @anthropic-ai/claude-code, needs Node.js 22 or newer. Anthropic’s install docs list every method. Confirm it with claude --version in a new terminal.
Your first project follows a short sequence. Open a terminal in the project folder and type claude. Sign in through the browser window. Run /init so it writes a starter CLAUDE.md. Switch to plan mode, describe the change, read the plan, and approve it. Then review the diff like a colleague’s work before you commit.
Install it only from Anthropic’s own docs. Fake install pages on lookalike domains have spread password-stealing malware, so type the docs address yourself.

Plan mode: the rule beginners break
The one rule beginners break is skipping plan mode. They describe a feature, it starts editing at once, and the result sprawls in ways nobody asked for. In plan mode, Claude reads the code, asks questions, and writes a plan, but edits nothing until you approve it.
Press Shift+Tab to cycle permission modes until you reach plan mode, or start with claude --permission-mode plan. You can also type /plan before a single prompt. Press Ctrl+G to edit the plan before anything runs.
Plan mode is overhead for tiny changes. Anthropic’s guidance is blunt: if the diff fits one sentence, skip the plan and just ask. Use it when a change touches several files or could break something you cannot see.

CLAUDE.md: project memory that survives every session
Every session starts empty, so it forgets your project each time it closes. CLAUDE.md is how the project rules carry over: a Markdown file read at the start of every session, holding build and test commands and conventions it cannot guess from the code.
Run /init in your first session and it writes a starter file from what it finds in the codebase. Then edit it by hand with the commands that must run and the rules that must hold. Keep it under 200 lines, since a bloated file eats context and gets followed less reliably. Cubic’s team guide covers CLAUDE.md and permission modes.
Commit the project CLAUDE.md with the code so the whole team shares it. A file at ~/.claude/CLAUDE.md holds your own preferences for every project. When it repeats a mistake, add the rule to the file.

Permission modes in plain words
Permission modes decide what the tool may do without asking first. Press Shift+Tab to cycle through them; the status bar shows the active mode.
Manual mode reads only: every edit and command waits for approval. acceptEdits runs reads, file edits, and common filesystem commands without asking, for work you review in the diff anyway. Plan mode reads and explores but makes no source edits. Auto mode runs most actions with a classifier checking each one, and recent versions start in it for terminal and VS Code sessions. Two stricter modes, dontAsk and bypassPermissions, exist for scripts and isolated machines only.
Start in Manual mode while you learn the tool. Move to accept-edits mode once you trust your review of the diff, and use auto mode for long tasks where prompts get tiring. Check the status bar before your first prompt, since the default has changed.

Pricing and where you can run it
Claude Code has no free plan and no price of its own: it is included in every paid Claude plan. Pro costs $20 a month, or $17 a month billed annually. Max costs $100 a month, Ultra costs $200 a month, and a Console account bills per token instead. Plans share one usage pool with Claude chat, with a five-hour window and a weekly limit. BenchClaw lists the current plan prices from Anthropic’s pricing page.
The same agent runs beyond the terminal. Extensions exist for VS Code and JetBrains IDEs, a desktop app offers visual diff review and parallel sessions, and the web version at claude.ai/code keeps working after you disconnect. Slack turns bug reports into pull requests when you mention @Claude in a channel.
Start in the terminal and build the plan-mode habit first. The ChatGPT models guide explains the rival lab’s model split, the DeepSeek AI guide covers a cheaper open-weight alternative, the Gemini AI Studio guide covers Google’s free workspace, and the best AI tools compared puts Claude Code in a wider field.

Frequently asked questions about Claude Code
What is Claude Code?
Claude Code is Anthropic’s agentic coding tool. It runs in your terminal, reads your codebase, plans changes, edits files, and runs tests.
Is Claude Code free?
No. You need Pro, Max, Ultra, Team, Enterprise, or a Console account that bills per token.
How do I install Claude Code?
Install it with the native installer, then type claude in your project folder and sign in. On macOS, Linux, or WSL the command is curl -fsSL https://claude.ai/install.sh | bash. On Windows PowerShell it is irm https://claude.ai/install.ps1 | iex.
What is plan mode in Claude Code?
Plan mode lets it explore your project and write a plan without editing source files. You approve the plan before any code changes.
What is CLAUDE.md?
CLAUDE.md is a Markdown file of project instructions read at the start of every session. Create a starter version with /init.
Which model does Claude Code use by default?
Opus 5.5 is the default on paid plans. Switch with /model, or start with claude --model sonnet for lighter work.