Anthropic · Free course

Learn Claude.

A free course on getting better results from Claude, from choosing a model and effort level to Projects, artifacts, connectors and Claude Code.

8 modules27 lessons108 techniques
Claude Code running inside the VS Code editor.
Syllabus

What's in this course.

Go in order, or jump to the lesson you need today. Each technique has a prompt you can copy.

  1. 01

    Basics

    Give Claude the context it needs, write clearer prompts, and choose the right model.

  2. 02

    Coding

    Use Claude and Claude Code to write, debug, test, review and restructure code in a real project.

  3. 03

    Writing

    Outline, draft, edit and document with Claude while keeping your voice and decisions consistent.

  4. 04

    Research

    Summarize documents, get citations, and compare options against criteria you set before Claude answers.

  5. 05

    Workflows

    Split work into linked prompts only when a step needs checking, and build a SaaS one working slice at a time.

  6. 06

    Hidden Features

    Learn artifacts, Projects, connectors and the effort control, and what each one is for.

  7. 07

    Advanced

    Choose what Claude sees, and design tools it can pick and use correctly.

  8. 08

    Mistakes

    Find out why Claude gives generic, agreeable or wrong answers, and how to fix the prompt and check the result.

Check yourself

How well do you know Claude?

Ten questions on using Claude well today, across prompts, code, research, connectors and writing. Each answer appears as you go, with a link to the lesson that explains it. No sign-up.

Take the quiz
How to use Claude well

Pick a model, then set the effort

Claude comes in four models: Fable, Opus, Sonnet and Haiku. Start with Sonnet for everyday work, move up when you can point to where an answer falls short, and move down to Haiku for a simple task you run many times.

Effort is the second control. On current models, Claude decides on each request how much to think, which Anthropic calls adaptive thinking. The older **Extended thinking** mode, with its fixed token budget, is now a legacy feature. Raise effort for multi-step problems where a wrong answer costs you, and lower it for quick work you can check. Higher levels take longer and reach your plan's limits faster.

Give Claude what it can't look up

Claude needs who the result is for, what has already happened, and what the result has to achieve. Anthropic's docs compare it to a brilliant new employee who doesn't know your norms, so tell it what you'd tell that person on day one. With long inputs, put the documents at the top and your question at the bottom. One real example can fix a format, and Anthropic recommends 3 to 5 for best results.

Paid plans give current models a 1M-token context window, roughly 555,000 words by Anthropic's estimate, but filling it usually isn't a good idea. Claude's built-in knowledge also stops at a cutoff, so turn on web search or paste the source for anything recent. Start a new chat when the topic changes or when Claude keeps returning to ideas you dropped.

Keep your work in one place

A project holds instructions and files for every chat inside it, and it suits one client or one product, not one task. An artifact opens a document, tool or dashboard in a panel beside the chat, so you change that piece instead of asking for a new version.

Connectors link Claude to your other services. The apps call them connectors, and they're built on MCP, the Model Context Protocol. Claude can do whatever the connected account can do, so limit the account, turn off tools you don't need, and approve write actions one at a time. An instruction in your prompt isn't a security control.

Use Claude Code with a check it can run

Claude Code reads your repo for itself, so CLAUDE.md should hold only what it can't work out: the commands that prove work is done, places where the code misleads, and anything you've corrected twice. Anthropic's docs say to target under 200 lines per file. Press Shift+Tab until plan mode is on when a wrong guess would be expensive, because edits stay blocked until you approve the plan.

Then give it a command that comes back pass or fail. Claude writes in the same fluent tone whether a detail is right or wrong, so judge the work by the command's output, not by a sentence saying it passed. Do the same outside code: ask for sources, and check any specifics you'll rely on.

Frequently asked
Is Claude free to use?

Yes. The Free plan includes Sonnet and Haiku, artifacts, up to five projects and one custom connector. It doesn't include Opus, Fable or Claude Code. Paid individual plans are Pro at $20 a month, Max at $100 and Ultra at $200, as listed in November 2026.

Which Claude model should I use?

Start with Sonnet for everyday work, and move up to Opus or Fable only when you can see the answer falling short. Move down to Haiku when you run the same simple task many times. The Free plan offers Sonnet and Haiku.

What happened to extended thinking in Claude?

It's now a legacy feature. On current models, Claude decides on each request whether and how deeply to think, which Anthropic calls adaptive thinking. The control you have is effort, with levels from low to max. You can expand the Thinking section above a response, but it shows a summary of Claude's approach, not its raw reasoning.

What is the difference between Claude connectors and MCP?

Connectors are what the Claude apps call their links to other services, and they're built on MCP, the Model Context Protocol. MCP is an open standard that lets AI apps reach tools and data through one interface. In the apps, you add a connector. In Claude Code, you run claude mcp add.

What is Claude Code and is it included in Claude Pro?

Claude Code is Anthropic's coding agent. It reads your repo, edits files and runs commands, and plan mode lets you approve its approach before any edits. Pro, Max, Ultra, Team and Enterprise include it, and its usage counts against the same limits as chat. The Free plan doesn't include it.

How long a document can Claude read?

Claude reads text and visuals in a PDF for up to 100 pages. From 101 to 1,000 pages it reads text only, and above 1,000 pages the upload fails. Split a long PDF if its charts matter. On paid plans, current models have a 1M-token context window, roughly 555,000 words by Anthropic's estimate.