Overview and setup
Claude Code is Anthropic's agentic coding tool — a command-line interface that gives Claude direct access to your local development environment. Unlike chatting with Claude in a browser, Claude Code can read your files, run commands, execute tests, and make edits directly in your codebase.
In the claude.ai chat interface, you and Claude exchange messages. You paste code in, Claude responds with suggestions, you copy them out and apply them yourself. Claude Code removes that friction entirely. You describe a task, and Claude reads the relevant files, reasons about the codebase, makes changes, runs tests, and iterates — all autonomously within your project.
This makes Claude Code fundamentally different in capability. It's not a smarter autocomplete; it's closer to a junior developer that can independently complete multi-step coding tasks.
Claude Code operates in a loop: it reads context, plans a course of action, takes a step (reads a file, runs a command, makes an edit), observes the result, and repeats. This loop continues until the task is complete or Claude needs input from you. Understanding this loop helps you work with Claude Code effectively — you're guiding an agent, not just asking questions.
Claude Code asks for permission before taking actions that could have significant consequences — deleting files, running scripts, making large changes. You control how much autonomy Claude has. For routine coding tasks, you can approve actions in bulk; for sensitive operations, Claude will always pause and ask.
