Claude 101 Study Guide: Complete Anthropic's Free Course and Earn the Certificate

Module 1, Meet Claude, and its four lessons: What is Claude?, Your first conversation with Claude, Getting better results and How you'll work with Claude on your desktop.

101_

Claude101

Anthropic's free course, and how to finish it

Register, work the five modules in order, claim the certificate. The hard part is never the lessons: it is the settings that silently hide half the features, the words that mean two different things depending on your plan, and the handful of places Anthropic's own documentation disagrees with itself. All of it is mapped below.

🎓

Free, start to finish

📚

5 modules, 14 lessons

Claude basics through connectors and research

🏆

Certificate of completion

No exam, no published passing score

✅

Fifteen practice questions

Ours, not Anthropic's, with worked answers

How to get through it

1. REGISTER

Free enrolment

anthropic.skilljar.com

The course page lists the whole curriculum before you sign up.

2. UNBLOCK

Turn the switch on

Settings > Capabilities

Artifacts and skills stay invisible until code execution is enabled.

3. PRE-READ

Skim the tab first

tabs 1 to 5

One tab per module, in the same order as the course.

4. PRACTISE

Build one of each

project, artifact, skill

The features only stick once you have made one yourself.

5. SELF-TEST

Fifteen questions

tab 5

Ours, not Anthropic's, with worked answers.

6. CLAIM IT

Certificate

Conclusion module

Finish the lessons, then add it to your profile.

Read the tab, do the lesson, build the thing

What the course coversIllustration

The five modules

Meet Claude4

Organizing your work3

Expanding Claude's reach3

Putting it all together2

Conclusion & certificate2

Cost

freeregister

Ends with

certificateno exam

module 2 › Creating with artifacts › what you actually need to know

Why can I not see the Artifacts section in my sidebar?

This study guideTAB 2

Artifacts need Code execution and file creation switched on first.
An artifact made inside a chat does not appear in the sidebar on its own.
You have to open it and click Publish to file it there.
On Team and Enterprise the button says Share, and it stays inside your org.
Unpublishing is permanent: that artifact can never be published again.

Where this comes from

[1]CourseModule 2, Creating with artifactslesson

[2]DocsWhat are artifacts and how do I use themhelp

[3]DocsPublish and share artifactshelp

[4]Guidetab 2, plus two self-test questionshere

every tab answers the questions the lessons skip past↓

What you will learn

Chat

Prompts that work

Projects

Reusable context

Artifacts

Shareable output

Skills

Repeatable procedure

Connectors

Your own tools

Research

Cited deep dives

What this page adds

The setting that hides artifacts and skills entirely
Publish versus share, and why unpublishing is forever
Why a skill you installed never fires
Connectors inherit your permissions, nothing more
The places Anthropic's own docs contradict each other
Fifteen practice questions with worked answers

Before you start

# 1. unblock the two features the course demos
Settings > Capabilities
  ☐ Code execution and file creation   # turn this on

# organisations do it here instead
Organization settings > Capabilities

# 2. check what you already have
Customize > Skills        # built-in skills listed?
Customize > Connectors    # anything connected?

# 3. then work modules 1 to 4, tab by tab

Why bother with the certificate

It is free and it is issued by the company that builds Claude. The real payoff is smaller and more useful: you stop using Claude as a chat box and start using projects, artifacts and skills, which is where the time saving actually lives.

5Modules

14Lessons

0Exams to pass

15Practice questions

8Tabs on this page

Turn this on before lesson one

One toggle blocks more of this course than anything else. With it off, artifacts and skills never appear at all.

Settings > Capabilities > Code execution and file creation

Team and Enterprise. The owner sets it at Organization settings > Capabilities, which sits above the personal switch.
It is on by default on every plan, so if it is off, someone turned it off.
It unlocks more than artifacts. XLSX uploads, and creation of Excel, PowerPoint, Word and PDF files.

Trap. File creation "consumes more limit tokens than standard conversations."

What Claude is

The definition. "A large language model (LLM) built by Anthropic," for writing, learning, summarization, programming, ideation, translation and image analysis.
Two beginner caveats. Models "may not be aware of events or information that occurred after their respective cutoff dates," and Claude can "hallucinate its capabilities," claiming it sent an email it cannot send. It has no access to tools "not explicitly integrated."

The second column is where new users lose an afternoon

Where Claude runs

Surface	Plans
`Chat` at claude.ai	`Free` and up
`Claude Desktop`	Chat on all plans, `Claude Code` and `Claude Cowork` on paid plans
`Claude for iOS`, `Claude for Android`	`Free` and up
`Claude in Chrome`	All paid plans, in beta
`Claude for Microsoft 365 (Excel, PowerPoint, and Word)`	All paid plans

Usage limits, in two clocks

Two clocks. A five-hour session limit that everyone has, Free included, plus a weekly limit on Pro, Max, Team and Enterprise that resets at a fixed time each week.
Usage is not length. "Usage limits control how much you can use Claude across all your conversations, while length limits control how long any single conversation can become."
One shared pool. claude.ai, Claude Code and Claude Desktop draw on the same limit.

Two clocks control how much you can use, a third caps how long one chat can get

What burns the limit faster: long messages sent often, big attachments, a long conversation as context accumulates, tools such as Research and web search, and file creation. Watch both clocks at Settings > Usage.

Getting better results

Talk normally. "Speak to Claude like you would a coworker or friend, naturally and conversationally."
The four starter tips. Anthropic lists exactly four: start simple, be specific ("the more specific your request, the better Claude can assist you"), iterate, explore.

Weak	Stronger	Why
"Write about our product launch."	"Write a 200 word launch announcement for existing customers, in flowing paragraphs rather than bullet points, because it goes into a newsletter that reads better as prose."	A constraint, an audience, a structure, then the reason, which lets Claude generalize the preference.

Anthropic's four starter tips, as one loop you actually run

No regenerate button. None is documented. To get a different answer, go back to a previous message and click Edit, which creates "a new conversation branch" you can always return from, or send a follow-up.

Memory remembers, when it is on. Claude builds it as entries it updates as you chat.
Each project has its own memory space and its own project summary.
Three controls. Toggle, pause (keeps existing memories, stops new ones), and reset, which permanently deletes all memories and is irreversible.

Next, module 2: projects, artifacts and skills.

Module 2, Organizing your work and knowledge, runs three lessons: Introduction to projects, Creating with artifacts and Working with skills.

Introduction to projects

What it is. "Projects allow you to create self-contained workspaces with their own chat histories and knowledge bases."
The rule of thumb. "Use projects for anything you'll reference multiple times."
Why it saves you limits. Project content is cached: "Every time you reference that content, only new/uncached portions count against your limits."
Its own memory. Each project has "its own separate memory space and dedicated project summary," kept apart from your non-project chats.
Free plan cap. "Free users can create a maximum of five projects."

What a project holds, and why the project name is not part of it

Two traps. The name is not context: "Give your project a name and description (note that Claude will not have access to these details)." And chats stay sealed off from each other: "Context is not shared across chats within a project unless the information is added into the project knowledge base." If a fact matters, upload it.

Creating with artifacts

What it is. "Claude can share substantial, standalone content with you in a dedicated window separate from the main conversation."
The six content types. Documents (Markdown or plain text), code snippets, single-page HTML websites, SVG images, diagrams and flowcharts, interactive React components.
Hard prerequisite. "We no longer support artifacts without Code execution and file creation enabled." Flip that switch first, back in tab 1.

The sidebar trap. "Artifacts you create in a conversation don't appear in your sidebar automatically. To add an artifact to your Artifacts section, open it and click Publish."

One button, two very different blast radii, decided entirely by your plan

Action	Plans	What it does
`Publish`	Free, Pro, Max	"Makes your artifact publicly available. Anyone with the link can view and interact with it."
`Share`	Team, Enterprise	"Makes your artifact available within your organization only." Org artifacts "cannot be published publicly."

Unpublishing is a one-way door. "Once you unpublish an artifact, you cannot publish that same artifact again," and it "permanently deletes all associated storage data." Treat Unpublish as delete, not pause.

Anthropic's pages disagree. The help center lists sidebar artifacts on every plan including Free. A claude.com tutorial says the tab exists only on Free, Pro and Max and "does not exist on mobile". This guide follows the help center.

Working with skills

What it is. "Skills are folders of instructions, scripts, and resources that Claude loads dynamically to improve performance on specialized tasks."
Progressive disclosure. "Claude determines which skills are relevant and loads the information it needs to complete that task, helping to prevent context window overload."
Four ship built in. Under Anthropic skills: Excel, Word, PowerPoint and PDF creation.
Upload your own. A folder holding at minimum the skill file, packaged as a ZIP with the folder as its root, uploaded at Customize > Skills.

How a skill actually fires, with no slash command anywhere in it

The myth to unlearn. There is no /skill-name command in ordinary chat. Slash invocation is documented only for the Microsoft 365 add-ins. Skills fire on their own: "You don't need to explicitly invoke them, Claude determines when each skill is needed based on your request." The description field is what decides: "Claude uses this to determine when to invoke your skill."

Check the toggle. Is the skill on in Customize > Skills?
Reread the description. "check the description clearly explains when it should be used".
Say it out loud. "Use my brand guidelines skill to create a presentation."

Anthropic's pages disagree. The help center puts skills on every plan, under Customize > Skills. A claude.com tutorial calls them a preview for paid plans at Settings > Capabilities > Skills. This guide follows the help center.

Projects hold your context, artifacts hold your output, skills hold your method. Tab 3 opens Claude's reach into the tools where your work already lives.

Module 3 is Expanding Claude's reach: Connecting your tools, Enterprise search and Research for deep dives.

Connecting your tools

Read this first. "Claude inherits each person's permissions from the connected service." If you cannot open a file, a channel or a record yourself, the connector cannot reach it either. A connector is a window onto what you can already see, not a key to what you cannot.

A connector is a window onto what you can already see, not a key to what you cannot

What a connector is. It lets Claude "access your apps and services, retrieve your data, and take actions within connected services". Directory connectors are built on the Model Context Protocol (MCP), the open standard underneath.
The first-party set: Google Drive, Gmail, Google Calendar, GitHub, Slack, Microsoft 365.

Open the directory from the + button in chat, or Customize > Connectors.
Click the connector and review its description and capabilities.
Click Connect or Install, then follow the authentication prompts.
Configure settings and permissions as needed.

A connector is a bundle of tools, governed one at a time under Tool permissions.

Setting	What it does
Always allow	Runs without asking. Only for a tool "that you trust to run unsupervised"
Needs approval	Claude asks first. "Review tool approval requests carefully"
Blocked	Off. Set per category or per individual permission

The three tool permission settings, from most to least trusting

Trap: the label is not a cage. Verified, Community and Custom describe the review a connector got before you connected it, not what it can do afterwards: "once connected, a community connector works the same way as a verified one." Your protection is the scopes and tool permissions you set, not the badge.

Trap: the Google consent screen. Google's screen mentions email sending permissions and people bail out there. "Claude only reads emails and creates drafts with your explicit approval", and "all emails must be sent manually through your Gmail account."

The named risk is prompt injection. "Malicious MCP servers may include hidden instructions that try to make Claude perform unintended actions." Claude has built-in protections, but protections are not a guarantee.

Enterprise search

What it is. "A dedicated project for searching across your organization's knowledge sources", a pre-built project sitting over several connectors at once.
Who gets it. "Available for users on Team and Enterprise plans."
Who sets it up. An Owner, once, picking connectors for Documents and Chat. Members then authenticate with their own credentials.

Research for deep dives

What it is. Claude searches "across both your internal work context and the web". It is not a search box: Claude "operates agentively, conducting multiple searches that build on each other".
What comes back. A report with inline citations that "link directly to the original material". Click them.
Plans and cost. Paid plans only (Pro, Max, Team or Enterprise), and no separate quota: "research sessions can use up your limits faster."

Trap: the prerequisite nobody mentions. "You must have web search turned on for research to function." If nothing happens, check web search, then ask for the tool by name.

Research is agentic, so it drains the same usage pool faster than a normal answer

Next: what each role actually does with all of this, on Putting it all together.

Module 4 is Putting it all together: Claude in action: use-cases by role and Other ways to work with Claude.

Claude in action: use-cases by role

Every example below is quoted from Anthropic's own pages.

Role	In Anthropic's words
Engineering	"Reproduces the bug, opens a draft PR, follows CI to green"
Sales	"Pull relevant CRM data, like details on comparable opportunities"
Marketing	"identify your best and worst performing channels"
HR	"Turn company information and a new hire's details into a personalized welcome guide"
Finance	"Create investment analyses with complete financial models"
Legal	"Review their changes against our playbook, flag every deviation"
Product	"Find recurring themes and pain points across user feedback"
Support	"Answers what it can, flags duplicates, routes the rest, rolls up themes"

One row of the table above, read as a workflow you could run this week

The use case library at claude.com/resources/use-cases, browsable by role.
The tutorials library at claude.com/resources/tutorials, filtered by feature.

Other ways to work with Claude

Claude Code, for developers. It works on your local files and you approve each change. Paid plans.
The Developer Platform and API at platform.claude.com, billed separately from any subscription.
Claude in Chrome, on all paid plans, in beta, Chrome desktop only.
Claude Cowork, a sandboxed background agent in Claude Desktop.
The Microsoft 365 add-ins for Excel, PowerPoint and Word, on all paid plans.

Trap. "The Pro plan does not include API usage through the Claude Console." The subscription and the API are separate products, billed separately, so paying for one buys you nothing on the other.

The distinction that saves an argument at renewal time

Next: test yourself on the self-test.

Module 5 is Conclusion & certificate, two lessons: What's next? and Certificate of completion. Here is the honest version of what that certificate is, and fifteen questions to check yourself first.

The certificate, honestly

There is no exam. The course ends with a lesson titled Certificate of completion. Not a test, not a graded assessment.
There is no published passing score. Anthropic does not publish a percentage, a quiz, or the exact completion requirements for this course. That is the state of the public record, not a gap in this guide.
So work the lessons. Register, open every lesson in order, finish module 5, and claim the certificate from its final lesson.
The self-test below is ours. The Testing Academy wrote these fifteen questions. They are not Anthropic's, they do not appear in the course, and passing them is not required for anything. They are a way to find out whether the features actually stuck.

The certificate path, start to finish

Worth having, and worth being honest about

Check yourself: fifteen questions

Weighted toward what learners actually get wrong in the product: the setting that hides two features, the word that means two things depending on your plan, and why a skill you installed never fires. Answer first, then reveal.

Q1 You are on a Pro plan. The Artifacts section is missing from your sidebar and the skills list is greyed out. What is the most likely cause?

A. Artifacts and skills are paid add-ons your plan does not include
B. You need Claude Desktop, because neither feature works on the web
C. Code execution and file creation is switched off in Settings > Capabilities
D. You have to ask Claude in the chat to enable them for your account

C. Both features hang off the same toggle. Anthropic states artifacts are no longer supported without code execution and file creation enabled, and the skills troubleshooting says a greyed-out skills section usually means code execution is disabled at the individual or organization level.

A is the tempting one, because it looks like plan gating. It is not: both features are documented on Free, Pro, Max, Team and Enterprise.

Q2 You built a chart artifact inside a chat yesterday. Today it is not listed in the Artifacts section of your sidebar. Why?

A. Artifacts are discarded once the conversation passes the context window
B. Artifacts created inside a conversation never appear in the sidebar automatically; open the artifact and click Publish to add it, or Share on Team and Enterprise
C. The sidebar only lists artifacts you started from the sidebar
D. The sidebar only lists artifacts created inside a project

B. Verbatim: "Artifacts you create in a conversation don't appear in your sidebar automatically. To add an artifact to your Artifacts section, open it and click Publish." On Team and Enterprise the equivalent button is Share, because those accounts cannot publish publicly at all.

C is close enough to feel right, since you can start an artifact from the sidebar, but publishing is the documented way to get an existing one listed.

Q3 You are on a Team plan and want a client outside your company to open an artifact you built. What does Claude let you do?

A. Click Publish and send them the public link
B. Click Share, then pick the anyone-with-the-link option
C. Nothing. Artifacts created on Team or Enterprise accounts can only be shared inside your organization and cannot be published publicly
D. Ask an Owner to publish it publicly on your behalf

C. The two words mean different things by plan. Publishing (Free, Pro, Max) "Makes your artifact publicly available. Anyone with the link can view and interact with it." Sharing (Team, Enterprise) "Makes your artifact available within your organization only. Viewers must be logged into your Team or Enterprise account to access it."

A is the trap, because Publish is the word most write-ups use for both.

Q4 You unpublish an artifact, then change your mind an hour later. What happens?

A. Click Publish again and the original link starts working
B. Republishing gives you a new link but keeps the stored data
C. You cannot publish that same artifact again. You have to create a new one, and any persistent storage data was permanently deleted
D. Anthropic support can restore the original link on request

C. Verbatim: "Once you unpublish an artifact, you cannot publish that same artifact again. You'll need to create a new artifact if you want to publish it later. Unpublishing also permanently deletes all associated storage data (both personal and shared) if the artifact used persistent storage."

Treat Unpublish as a one-way door, not a visibility toggle.

Q5 You uploaded a custom skill, but Claude ignores it on a task it obviously fits. What does Anthropic tell you to check?

A. Whether you typed /your-skill-name at the start of the message
B. Whether the skill is toggled on in the skills list, and whether its description clearly explains when it should be used
C. Whether the skill was added to the project's knowledge base
D. Whether you are on a paid plan, since custom skills are paid only

B. The description is what drives invocation: "This is critical, Claude uses this to determine when to invoke your skill." The documented fixes are to verify the toggle, sharpen the description and instructions, and be more explicit in the request, for example "Use my brand guidelines skill to create a presentation."

A is the popular wrong answer. Slash invocation of a skill is documented only for the Microsoft 365 add-ins, not for ordinary chat. In chat, skills fire on their own or not at all.

Q6 When does Anthropic's own guidance say to move work into a project instead of a plain chat?

A. Whenever a conversation runs past twenty messages
B. Only when you need to share the work with your organization
C. For anything you will reference multiple times, because project files are cached and reused without burning your usage as fast
D. Only when you need to attach files, since plain chats cannot take attachments

C. Verbatim: "Use projects for anything you'll reference multiple times." The economics are the reason: "When you upload documents to a project, they're cached for future use. Every time you reference that content, only new/uncached portions count against your limits."

D is simply false: a plain chat accepts up to 20 files.

Q7 Two chats sit inside the same project. Chat one contains a decision you want chat two to know about. Does chat two see it?

A. Yes. Everything inside a project is shared context
B. Yes, as long as both chats stay inside the project
C. No. Context is not shared across chats within a project unless the information is added into the project knowledge base
D. No, and project instructions do not carry across chats either

C. Anthropic states it as a note precisely because everyone assumes otherwise: "Context is not shared across chats within a project unless the information is added into the project knowledge base." If a decision matters, write it into project knowledge.

D goes too far the other way. Project instructions and project knowledge do apply to every chat in the project. It is the chats that cannot see each other.

Q8 You want Claude to keep everything it already knows about your work, but stop learning anything new. Which control does that?

A. Reset memory
B. Pause memory
C. Do all your work in incognito chats from now on
D. Delete your chat history

B. Pause keeps existing memories and stops new ones being created. The three documented controls are toggle memory on or off, pause, and reset.

A is the dangerous one: reset permanently deletes all memories and is irreversible. C only excludes the individual conversations you run that way, since incognito chats are not saved to history or to memory.

Q9 Claude replies "Your message will exceed the length limit for this chat." What has actually happened?

A. You hit your five-hour usage limit and have to wait for the reset
B. You hit your weekly limit
C. You hit the length limit of this single conversation. Break the content into chunks, summarize first, or start a new chat
D. Anthropic is over capacity, so wait a few minutes and retry

C. Two different clocks, and beginners conflate them: "Usage limits control how much you can use Claude across all your conversations, while length limits control how long any single conversation can become."

D is a real error message, but a different one, and it names capacity explicitly.

Q10 You do not like an answer and want Claude to try again differently. What does Anthropic actually document?

A. A regenerate icon underneath the reply
B. A thumbs down, which triggers a retry
C. Click Edit on an earlier message to create a new conversation branch you can always return from, or send a follow-up
D. Switching the model, which re-answers the message you already sent

C. Verbatim: "Go back to any previous message, click Edit to create a new conversation branch, and try different approaches. You can always return to your original version."

D is half true and fully misleading: a model change takes effect from Claude's next response, so it does not redo the answer already on screen. No regenerate button or thumbs rating is documented for a Claude chat reply.

Q11 A teammate can open a Drive folder that you cannot. You connect the Google Drive connector and ask Claude about that folder. What happens?

A. Claude reads it, because the connector authenticates at the organization level
B. Claude reads it, because an Owner approved the connector for the whole team
C. Claude cannot reach it. Claude inherits each person's permissions from the connected service
D. Claude reads it but redacts the parts you are not cleared for

C. The rule is stated repeatedly: "Claude inherits each person's permissions from the connected service", and "you can only sync content to Claude that you have permission to view in the original source."

B is the trap. On Team and Enterprise an Owner enabling a connector only makes it available. Each member still authenticates with their own credentials and still sees only their own data.

Q12 A connector in the directory carries a Community label. What does that tell you about how it behaves after you connect it?

A. It runs with fewer permissions than a verified connector
B. It cannot take write actions until Anthropic reviews it
C. Nothing. It runs the same way a verified connector does, on the same infrastructure. The label describes the review it got before you connected, not what it can do afterwards
D. Anthropic reviewed it in depth, it was just built by someone else

C. Verbatim: "The label is a quality signal. It affects how the connector is displayed and discovered in the directory, not how the connector itself functions: once connected, a community connector works the same way as a verified one." And: "Directory connectors and custom connectors run on the same infrastructure."

D inverts the definition. Community means it "passed Anthropic's automated checks, but Anthropic has not reviewed it in depth." Your protection after connecting is the scopes you granted and the tool permissions you set.

Q13 While connecting Gmail, Google's consent screen mentions email sending permissions. What can Claude actually do with your mail?

A. Send messages on your behalf once you approve each one
B. Read emails and create drafts. It cannot send, so every email goes out manually from your Gmail account
C. Nothing at all until an administrator grants a tenant-level consent
D. Read, send and delete, which is exactly why the screen asks for it

B. Anthropic pre-empts this exact moment: "During authentication, Google's OAuth screen mentions email sending permissions. Claude only reads emails and creates drafts with your explicit approval", and "all emails must be sent manually through your Gmail account."

C describes the Microsoft 365 connector, where a Microsoft Entra Global Administrator grants a one-time tenant-level consent. It does not describe Google.

Q14 You switch on Research, ask a big question, and get an ordinary short answer. What is the documented prerequisite to check first?

A. Extended thinking has to be on
B. Web search has to be turned on, or research does not function
C. You have to be working inside a project
D. At least one connector has to be attached

B. Verbatim: "You must have web search turned on for research to function." The second thing to try is asking for the tool by name, since Anthropic notes it may not trigger on its own.

A is tempting because Anthropic does recommend pairing extended thinking with research so Claude can "both plan its approach thoughtfully and execute comprehensive information gathering." Recommended is not required.

Q15 What does running Research do to your usage?

A. Nothing. Research has its own separate quota
B. It is billed separately, at API rates
C. It draws on the same limits as normal conversations and uses them up faster, and it is only available on paid plans
D. It is free to run on every plan, including Free

C. Verbatim: "Research is subject to the same limits as standard Claude conversations. However, research sessions can use up your limits faster due to Claude retrieving multiple sources and providing comprehensive responses." And: "Research is available for users with paid Claude plans (Pro, Max, Team, or Enterprise)."

D is wrong twice over: it is not on Free, and nothing about it is free of usage cost.

What to study next

Anthropic's use case library, for worked examples across every category, each with the same four-part structure you can copy as a prompt template.
Anthropic's tutorials library, filtered by feature rather than role. Take Skills, Connectors, Projects and Research in that order and you have re-taken modules 2 and 3 with hands-on material.
The help center at support.claude.com, which is where every fact on this page came from. If any tutorial disagrees with it, believe the help center.

The fastest way to make this stick. Take one repeated task from your week and rebuild it three ways: as a well-written prompt, then as a project holding your reference documents, then as a skill that fires on its own. By the third pass you will understand the difference better than any lesson can teach it, and you will have something to point at when someone asks what the certificate means.

That is the whole course. Go and finish it: anthropic.skilljar.com/claude-101.