What Is Claude Code? What It Does and How to Start Using It

Claude Code is Anthropic’s agentic coding tool. It runs in your terminal, reads your project, and can edit files, run commands, and work through multi-step programming tasks on its own. You sign in with a Claude account or use an API key, describe what you want, and it does the work while you watch and approve.

That makes it different from a chatbot that hands you code snippets to copy. Claude Code works directly inside your codebase.

What Claude Code actually does

At its core, Claude Code is an agent: a model that can take actions in a loop until a goal is reached. If you are new to the idea, our explainer on what an AI agent is covers the basics.

In practice, a typical session looks like this:

• You open a terminal in your project folder and start Claude Code.
• You describe a task, such as “add a dark mode toggle to the settings page.”
• It reads the relevant files to understand how your project is structured.
• It proposes and makes changes across one or more files.
• It runs commands, like your tests or build, to check its own work.
• It reports back what it changed, and you review it.

The key word is context. Because it can open and search your files, it doesn’t need you to paste everything into a chat window. It finds what it needs.

Things it is good at

Claude Code tends to shine on tasks that require understanding how pieces of a project fit together:

• Explaining unfamiliar code. Ask “how does authentication work in this repo?” and it will trace through the files.
• Multi-file changes. Renaming a concept, adding a feature that touches the UI and the backend, or updating an API and its callers.
• Fixing bugs. Give it an error message and it will look for the cause, try a fix, and rerun the failing step.
• Writing tests. It can read existing tests, match their style, and add coverage.
• Routine chores. Updating configuration, cleaning up warnings, writing a README, or preparing a commit message.

Things to keep an eye on

Like any coding agent, it can be confidently wrong. It may misread a requirement, take a shortcut, or change more than you expected. The fix is not to avoid it, but to give it clear goals and checks it can run, and to review its work like you would review a colleague’s.

How Claude Code works under the hood (conceptually)

You don’t need to know the internals to use it, but a mental model helps.

1. You give a goal. Plain language is fine.
2. The model plans. It decides what to look at first.
3. It uses tools. Reading files, searching, editing, and running shell commands are all “tools” the model can call.
4. It observes results. Command output, test failures, and file contents feed back into its next step.
5. It repeats until it believes the task is done, then summarizes.

This loop is what people mean by an agentic workflow. The model is not just generating text; it is acting and reacting.

Permissions

Because it can run commands on your machine, Claude Code asks for permission before doing things that could have side effects. You can approve actions one at a time or allow certain kinds of actions in advance. Starting cautious is a good habit: let it read freely, but review edits and commands until you trust how it behaves on your project.

How to start using Claude Code

Getting started is straightforward. The exact install steps live in Anthropic’s official documentation, so check there for the current command. The general path looks like this:

1. Install it following Anthropic’s instructions for your operating system.
2. Sign in with your Claude account, or configure an API key.
3. Open a terminal in a project folder. Start with a small or non-critical project.
4. Start Claude Code and ask it to explain the project. This is a safe first task that shows you how it explores code.
5. Give it a small, concrete change. For example, “add input validation to the signup form and show an error message under the email field.”
6. Review the diff. Read what changed before you keep it. Use version control so you can always roll back.

Tips for your first week

• Commit before you start a task. A clean git state makes it easy to see exactly what the agent changed.
• Tell it how to verify. “Run the tests after your change” is one of the most useful sentences you can add.
• Keep tasks scoped. One feature or one bug at a time works better than “improve the app.”
• Give it project notes. Many teams keep a short file of conventions, commands, and rules that the agent reads at the start of each session. It saves you from repeating yourself.
• Ask it to plan first. For bigger tasks, ask for a plan, adjust it, then let it implement.

If you want to go deeper on writing good instructions, read how to write a brief an AI coding agent can actually finish.

Who is Claude Code for?

It is built for developers, but the audience is wider than that.

• Professional developers use it to move faster on routine work and to explore unfamiliar codebases.
• Solo builders and indie hackers use it to ship features they would otherwise not have time for.
• Technical non-programmers, like designers or product people comfortable with a terminal, use it to build prototypes and internal tools.

You still need to judge the result. The more you understand what “done” should look like, the better your outcomes.

Claude Code compared to other coding agents

Claude Code is one of several terminal coding agents. OpenAI has Codex, Google has the open-source Gemini CLI, and OpenCode is an open-source agent that can use many model providers. They share the same basic idea, with different models, defaults, and sign-in options.

Tool	Made by	Sign in with
Claude Code	Anthropic	Claude account or API key
Codex	OpenAI	ChatGPT account or API key
Gemini CLI	Google	Google account or API key

For a fuller side-by-side, see Claude Code vs Codex vs Gemini CLI.

Running more than one Claude Code at a time

Once you are comfortable with a single agent, a natural next step is running several in parallel: one on the frontend, one on tests, one reviewing. That works, but coordinating them by hand across terminal windows gets messy quickly.

That is the problem BotQuarters is built for. It runs a small team of coding agents on your Mac, where a team lead splits your goal into tasks, assigns teammates (which can use Claude Code alongside other agents), reviews their work, and runs your project’s checks before anything is merged. You bring your own Claude account.

FAQ

Is Claude Code free?

Claude Code itself is a tool from Anthropic, but using it requires access to Claude, either through a Claude account plan that includes it or through an API key that is billed by usage. Check Anthropic’s site for the current plans and what each includes.

Do I need to know how to code to use Claude Code?

Not strictly, but it helps a lot. You need to be comfortable opening a terminal and describing what you want clearly. Being able to read code, even roughly, makes it much easier to review the changes and catch mistakes.

Is Claude Code safe to run on my computer?

It asks for permission before taking actions with side effects, and you control what it is allowed to do automatically. Good practice is to work in a project under version control, review edits, and avoid granting broad permissions until you trust its behavior on your project.

What is the difference between Claude Code and the Claude chat app?

The chat app is a conversation: you paste code in and copy answers out. Claude Code works directly in your project, reading files, making edits, and running commands itself. It is built for doing the work, not just discussing it.