# Claude 101: Foundations for Everyday Work

> Claude 101 is the on-ramp course that teaches you what Claude is, how to write prompts that land, and how to organize work using Projects, Artifacts, Skills, and Connectors. It is a product tour, not an architecture course, but it grounds every later concept in the actual UI you will be tested on. If you already use claude.ai daily you can skim it; if not, do not skip it.

**Domain:** D5 · Context + Reliability (15%)
**Difficulty:** intro
**Skilljar course:** Claude 101 (14 lessons)
**Canonical:** https://claudearchitectcertification.com/knowledge/claude-101
**Last reviewed:** 2026-05-06

## Exam mapping

**Blueprint share:** 15% (D5)

Direct prep for D5 task statements on the Claude product surface (Chat, Projects, Artifacts, Skills, Connectors) and the human-in-the-loop habits the exam frames as evaluation and iteration. Light overlap with D2 prompt clarity practices.

## What you'll learn

- What Claude is, what Constitutional AI means in plain language, and where Claude is available (web, desktop, Slack, Excel)
- How to write prompts that beat the five canonical failure modes (too generic, wrong length, wrong format, hallucinated facts, wrong tone)
- How Projects (knowledge), Artifacts (output), Skills (procedure), and Connectors (tools) divide responsibility cleanly
- When to reach for Skills vs Projects (procedure vs knowledge) without overlapping the two
- How the 4D Framework (Delegation, Description, Discernment, Diligence) frames every later course

## Lesson outline

### 1. What is Claude?

Claude is a steerable AI assistant trained under Constitutional AI; available on web, desktop, mobile, Slack, and Excel.

### 2. Your first conversation with Claude

Walk through the chat UI, message history, and the basic conversational rhythm of asking and refining.

### 3. Getting better results

Five canonical failure modes (generic, length, format, hallucination, tone) with the iteration mindset and a simple eval recipe.

### 4. Navigating the Claude desktop app: Chat, Cowork, Code

Three surfaces in one app; Chat for conversation, Cowork for delegation on real files, Code for agentic dev work.

### 5. Introduction to projects

Self-contained workspaces with knowledge, instructions, and chat history; auto-RAG when knowledge exceeds context.

### 6. Creating with artifacts

Standalone interactive outputs (docs, code, HTML, SVG, Mermaid, React) rendered alongside the chat for reuse and sharing.

### 7. Working with skills

Folders of instructions and scripts Claude loads dynamically; Anthropic-built or custom; procedure not knowledge.

### 8. Connecting your tools

MCP-powered web connectors (Drive, Slack, Linear, Stripe) and desktop extensions; Claude only sees what you can see.

### 9. Enterprise search

Claude for Work feature that connects Claude to your org's knowledge sources with custom prompts.

### 10. Research mode for deep dives

Long-running research that searches, reads, and synthesizes citations into a structured report.

### 11. Claude in action: use-cases by role

A use-case gallery walkthrough; sales, marketing, finance, HR, legal, research patterns.

### 12. Other ways to work with Claude

Pointer to Claude Code, Claude in Chrome, Claude in Excel, Claude in Slack as specialized surfaces.

### 13. What you've learned

Recap of the four organizing primitives and the iteration mindset.

### 14. Certificate of completion

Skilljar certificate; not exam-relevant on its own but a signal of foundations coverage.

## Our simplification

Claude 101 looks like a product tour but it is actually the course that teaches you the four primitives every later course assumes you know. Claude is not a chatbot; it is a steerable assistant trained under Constitutional AI, available across web, desktop, mobile, Slack, and Excel. The course frames Claude as a thinking partner, which sounds soft but matters on the exam because every D5 question about appropriate use, evaluation, and human-in-the-loop traces back to this framing. Skip the lesson and you miss the vocabulary the rest of the curriculum uses.

The load-bearing lesson is Lesson 3, Getting better results. It catalogs the five failure modes you will hit on day one: responses too generic, wrong length, wrong format, confidently wrong facts, wrong tone. Each has a fix that is mechanical, not magical: add audience and constraints, ask for explicit length, show a format example, ask for sources, describe tone in plain language. The lesson also introduces the iteration mindset (treat first drafts as starting points, give specific feedback, know when to start fresh) and the 4D Framework (Delegation, Description, Discernment, Diligence). The 4D vocabulary reappears across every Anthropic course, and the iteration habits are the practical answer when D5 exam questions ask how a responsible operator should use Claude. Learn this lesson once, deeply, and you save time on every later course.

Projects are the organizing primitive for knowledge. A project is a self-contained workspace with its own chat history, knowledge base, and instructions. Project instructions apply to every conversation in the project; this is how you stop re-explaining tone, audience, and constraints turn after turn. When the knowledge base exceeds the context window, Claude transparently switches to RAG mode (search-and-retrieve) so the project capacity expands roughly 10x. For Claude for Work users, projects also enable team collaboration with view, edit, and owner roles.

Artifacts are the organizing primitive for output. When Claude generates something significant and self-contained (typically over 15 lines, something you would edit, iterate on, or reuse) it renders it in a dedicated panel beside the chat instead of dumping it inline. Artifacts cover documents, code, HTML pages, SVGs, Mermaid diagrams, and live React components. The interaction model is iterate incrementally: one change at a time so you can see what each instruction did. Free, Pro, and Max users can publish artifacts to a public link; Team and Enterprise users share them inside the org.

Skills vs Projects is the single most-confused distinction in the course, and it lands on the exam in disguise. The clean rule: projects store knowledge, skills perform tasks. A project holds the brand guidelines PDF; a skill encodes the procedure for applying them to every deck you generate. Skills are folders of instructions and scripts Claude loads dynamically; Anthropic ships built-in skills for Excel, Word, PowerPoint, and PDF creation, and you can author custom skills by chatting with Claude itself. Skills require code execution and file creation to be enabled.

Connectors are the organizing primitive for reach. They are powered by MCP (Model Context Protocol), which the course memorably describes as USB-C for AI: a universal standard that lets any developer build a connector and any Claude surface use it. There are two kinds: web connectors (Drive, Notion, Slack, Asana, Linear, Stripe, and many others) and desktop extensions (local file access, browser control, native app integration via the Claude Desktop app). The safety contract is simple and worth memorizing: Claude only sees what you see. Connecting your work email never grants Claude access to anyone else's inbox, every permission is scoped to what the connector actually needs, and every grant is revocable from either Claude's settings or the third-party service. For enterprise environments, Enterprise Search (Lesson 9) layers custom prompts on top of organization-specific knowledge sources.

If you already use claude.ai daily, you can skim Claude 101 for the vocabulary and move on. If you do not, do not skip it; later courses will assume you know what an artifact is, what a connector does, and why projects exist. Treat Lessons 3, 5, 6, 7, and 8 as the load-bearing five; the rest are useful but not exam-critical. The 4D Framework, the projects-vs-skills split, and Claude only sees what you see are the three things you should be able to recite on demand.

## Patterns

### 5 canonical prompt failure modes (and the fix for each)

Lesson 3 of Claude 101 catalogs the five most common day-one failures. Memorize the fix, not the failure.

- **Response is too generic.** Your prompt didn't include enough context. Fix: add audience, role, and constraints. Write an email about the delay becomes Write an email to our enterprise client explaining the integration is delayed two weeks; this is the second delay; keep it professional but apologetic.
- **Response is the wrong length.** Claude is guessing. Be explicit: two-paragraph summary, under 100 words, comprehensive analysis, length isn't a concern. Length is a parameter, not a hint.
- **Format isn't what you wanted.** Claude understood the what but not the how. Fix: show, don't tell. Provide a format example or describe structure: bullet points with bold headers per section.
- **Confidently wrong facts.** Hallucinations are most likely on niche specifics. Fix: ask Claude to cite sources or indicate confidence; for high-stakes work, verify independently; enable web search to ground in current information.
- **Tone isn't right.** Claude defaults to helpful and professional. Fix: describe tone in plain language (more conversational, authoritative and formal) and provide an example of writing in the style you want.

### Projects vs Skills, in one sentence each

The cleanest mental model from Lesson 7. If you can recite this, you've already passed half the D5 questions on the surface.

- **Projects store knowledge.** Long-term reference materials, persistent context across all chats in the project, team-collaboration target. Use for client hubs, research repositories, brand kits.
- **Skills perform tasks.** Procedural how-to packages Claude loads when relevant. Use for repeatable workflows: brand-voice review, quarterly variance analysis, deck templating.
- **They compose.** A customer call prep skill can pull from customer profiles in a project's knowledge base. The project provides the what (information); the skill provides the how (process).

## Key takeaways

- Claude is built around four primitives: Projects (knowledge), Artifacts (output), Skills (procedure), Connectors (reach). Every later course inherits this vocabulary. (`skills`)
- Project instructions apply to every chat in the project; when knowledge exceeds the context window, Claude transparently switches to RAG mode for ~10x capacity. (`context-window`)
- Artifacts trigger automatically for content that's substantial, self-contained, and reusable; you can also explicitly request Create this as an artifact if Claude misses the cue. (`structured-outputs`)
- Skills are procedural (how Claude executes); Projects are referential (what Claude knows). They compose, not overlap. (`skills`)
- Connectors are powered by MCP and follow the rule Claude only sees what you see; permissions are scoped, revocable, and per-service. (`mcp`)
- The 4D Framework (Delegation, Description, Discernment, Diligence) is the AI-fluency vocabulary every later Anthropic course assumes; learn it here. (`4d-framework`)

## Concepts in play

- **Skills** (`skills`), One of the four product primitives
- **Model Context Protocol** (`mcp`), How connectors are wired
- **4D Framework** (`4d-framework`), AI-fluency vocabulary used across every Anthropic course
- **Context window** (`context-window`), Why Projects auto-switch to RAG mode
- **System prompts** (`system-prompts`), What Project Instructions actually are

## Scenarios in play

- **Claude for operations** (`claude-for-operations`), Practical use of Projects + Connectors for ops-team knowledge work
- **Agent Skills for enterprise KM** (`agent-skills-for-enterprise-km`), Where the Projects-vs-Skills split shows up at enterprise scale

## Curated sources

- **What are Skills? Anthropic Help Center** (anthropic-blog, 2025-10-15): Canonical reference for the Skills feature; pairs with Lesson 7 when you start building custom skills and need the exact enable-step sequence.
- **How can I create and manage projects? Anthropic Help Center** (anthropic-blog, 2025-08-01): The walkthrough Lesson 5 abbreviates; useful when you want the screenshots, sharing-permission matrix, and current pricing-tier gating.
- **Introducing the Model Context Protocol** (anthropic-blog, 2024-11-25): Anthropic's announcement of MCP itself; deeper than Lesson 8's USB-C for AI analogy and required reading before MCP Foundations course.

## FAQ

### Q1. What is Claude 101 and is it worth taking before the Claude Architect exam?

Claude 101 is Anthropic's free 14-lesson Skilljar course that introduces the Claude product surface: Chat, Projects, Artifacts, Skills, and Connectors. It's the only course that teaches the vocabulary every later course assumes you know, so yes, take it first if you don't use claude.ai daily. If you do, skim Lessons 3, 5, 6, 7, 8 and move on.

### Q2. What's the difference between Claude Projects and Claude Skills?

Projects store knowledge; Skills perform tasks. A Project is a workspace with persistent knowledge, instructions, and chat history Claude references across every conversation in the project. A Skill is a procedural package (instructions, scripts, resources) Claude loads when it determines the workflow applies. They compose: a skill can read knowledge from a project's knowledge base.

### Q3. When does Claude create an artifact vs reply inline in chat?

Claude creates an artifact when content is significant and self-contained (typically over 15 lines), is something you'll likely edit or reuse, and stands on its own without the surrounding conversation. If Claude replies inline when you expected an artifact, just ask: Create this as an artifact.

### Q4. How do I get Claude to follow a specific format every time?

Show, don't just tell. Provide an example of the format directly in the prompt or in your project's Instructions. Vague tells (use bullet points) work less reliably than examples. For repeatable workflows that need the same format every time, encode the procedure in a custom Skill so Claude follows the steps consistently.

### Q5. What can Claude actually access through connectors and is it safe?

Connectors are powered by MCP (Model Context Protocol) and grant Claude scoped, revocable access to your tools (Drive, Slack, Linear, Stripe, and others). The safety rule is Claude only sees what you see: connecting your work email gives Claude access to your inbox, never anyone else's. Permissions are revocable from Claude's settings or the third-party service at any time.

### Q6. Why does Claude give me confidently wrong information sometimes?

Claude occasionally generates plausible but incorrect content, especially on niche specifics or recent events. The fixes are mechanical: ask Claude to cite sources, ask it to indicate confidence level, enable web search to ground responses in current information, and for high-stakes work verify key facts independently. Treat first drafts as starting points, never as final answers.

### Q7. Do I need a paid Claude plan to use Projects, Artifacts, Skills, and Connectors?

Projects and Artifacts are available on Free, Pro, Max, Team, and Enterprise. Skills require Pro, Max, Team, or Enterprise (and code execution + file creation enabled). Connectors are available across plans, but enterprise-grade connectors and higher rate limits are gated to paid tiers. Check your plan's Capabilities settings.

### Q8. What is the 4D Framework and why does it matter for the certification?

The 4D Framework is Anthropic's AI-fluency vocabulary: Delegation (deciding human vs AI work), Description (clearly communicating with AI), Discernment (critically evaluating AI output), Diligence (using AI responsibly). Every later Anthropic course assumes this vocabulary, and exam questions on appropriate use, evaluation, and oversight trace back to it. Learn it once in Claude 101 and you don't have to relearn it.

---

**Source:** https://claudearchitectcertification.com/knowledge/claude-101
**Vault sources:** Course_01_claude-101/_Course_Overview.md; Course_01_claude-101/Lesson_01_what-is-claude.md; Course_01_claude-101/Lesson_03_getting-better-results.md; Course_01_claude-101/Lesson_05_introduction-to-projects.md; Course_01_claude-101/Lesson_06_creating-with-artifacts.md; Course_01_claude-101/Lesson_07_working-with-skills.md; Course_01_claude-101/Lesson_08_connecting-your-tools.md; Course_01_claude-101/Lesson_11_claude-in-action-use-cases-by-role.md
**Last reviewed:** 2026-05-06