Plain-words intro: this level teaches four things — ① five small tricks to phrase requests clearly ② drop in files and get the key points ③ web research with sources you can check ④ build a clickable mini tool just by describing it (Artifacts). No tech background needed.
L1 · Foundations
Audience: complete beginners | Suggested duration: 3–4 hours | Focus: Claude Chat fundamentals, the prompting framework, Artifacts, and safety awareness Maps to the
claude-courseoutline: Meet Claude, Your First Conversation, and Prompting Essentials — the fastest “day one” on-ramp
1. Learning Objectives (by the end of this level, you can…)
- Write clear prompts using the Five-Element Prompt Framework (Role + Task + Context + Format + Constraints).
- Build an interactive, shareable mini-tool with Artifacts from a single sentence.
- Upload a file or PDF and have Claude summarize it and pull out the key numbers.
- Use web search to get up-to-date information — and click through the citations to verify.
- Know the safety baseline: AI can hallucinate, so verify; never upload sensitive data — or use an Incognito chat.
2. Core Modules
| Module | Concept | Why it matters |
|---|---|---|
| The Three Modes | How Claude Chat, Cowork, and Claude Code differ | Builds the mental model so nothing gets confusing later |
| Five-Element Prompt Framework | Role · Task · Context · Format · Constraints | Prompt quality determines output quality (the foundation of the whole course) |
| Artifacts | One sentence → an interactive tool | The lowest-barrier “wow” — build a tool with zero code |
| Files / multimodal | Read PDFs, images, and tables | Turn your existing documents into something you can ask questions of |
| Web search + verification | Live search with citations | Breaks the “knowledge cutoff” myth and builds verification discipline |
| Safety awareness | Hallucination / privacy / Incognito chat | The baseline for using AI safely, for the long haul |
3. Demo Packs (4 in this level; = pre-built example, = live operation with a runbook)
| # | Demo | Type | Source file (zh-TW course repo) |
|---|---|---|---|
| D1 | Artifacts: build an interactive quiz in one sentence | Pre-built HTML example | 課程/L1_基礎啟動/demo包/D1_Artifacts互動測驗/ |
| D2 | Five-Element Prompt Framework (before/after comparison) | Example prompt + output | 課程/L1_基礎啟動/demo包/D2_提問框架五要素.md |
| D3 | Upload a PDF: summarize + extract the numbers | Requires a live upload | 課程/L1_基礎啟動/demo包/D3_檔案摘要_runbook.md |
| D4 | Web search + citation verification | Requires a live search | 課程/L1_基礎啟動/demo包/D4_Web搜尋查證_runbook.md |
Demos that require live instructor operation are consolidated in the course’s pending-actions checklist.
4. Four-Track Hands-On Labs (one skill, four role scenarios)
Office & Admin — Turn meeting notes into a to-do list
-
Goal: turn long, messy meeting notes into 5 key takeaways plus an action-item table with owners and deadlines.
-
Prompt:
You are a project assistant. Organize the meeting notes below into: (1) the 5 key takeaways, and (2) an action-item table with columns for Task / Owner / Deadline. Meeting notes: [paste here]
-
Pass criteria: output has clear takeaways plus a tabulated action list; no decisions from the meeting are missing.
Sales — Prospect icebreakers
-
Goal: research a prospective customer and generate 3 icebreaker openers.
-
Prompt (turn on web search):
You are a B2B sales rep. Search the web for recent news about [company name]. Based on what they do, give me 3 professional, non-salesy icebreaker openers, each with the reasoning behind it.
-
Pass criteria: each opener cites something specific and recent about the company, with checkable citations.
Nonprofit Advocacy — One-page policy explainer
-
Goal: turn a policy or bill PDF into a one-page issue explainer.
-
Prompt:
You are a policy staffer. Summarize this PDF into: (1) the issue in one sentence, (2) the 3 key points of contention, and (3) the impact on the communities we serve. Cite the page number for every claim.
-
Pass criteria: every point in the summary traces back to a page number; the points of contention are stated neutrally, without taking sides.
Founders — Market research tool
-
Goal: use Artifacts to build a shareable market-research survey or calculator.
-
Prompt:
Build an interactive market-research survey (5 questions covering industry, pain points, and budget). When someone finishes, show a chart of the response distribution. Make it shareable.
-
Pass criteria: an answerable, shareable survey appears in the Artifacts panel.
5. Anti-Patterns / Common Mistakes
- One vague sentence (“write me something”) → use the Five-Element Prompt Framework to add context.
- Treating AI like a search engine and never verifying → turn on web search and click the citations.
- Asking for a tool but getting a wall of text → say explicitly: “build this as an interactive Artifact.”
- Uploading sensitive personal data or case files → switch to an Incognito chat, or don’t upload at all.
- Cramming too many tasks into one prompt → break it into small steps and ask one at a time.
- Blindly trusting numbers → manually re-check any figure that matters (see the safety module).
6. Capstone Project
Build “my first AI tool + an AI-assisted summary report”:
- Use Artifacts to build a small tool related to your own work (a survey, calculator, or quiz), and produce a share link.
- Use file upload + web search to produce a one-page summary report — with verified numbers and cited sources.
7. Competency Rubric
| Competency | Not yet | Proficient | Mastery |
|---|---|---|---|
| Prompting framework | Still throws in one vague sentence | Uses all five elements | Adapts the framework flexibly to the task |
| Artifacts | Only gets text output | Can build one interactive tool | Can iterate on it and share it with others |
| Files / verification | Uses output without checking | Summarizes + clicks citations | Has internalized “AI gathers, humans verify” |
| Safety awareness | Doesn’t distinguish sensitive data | Knows not to upload sensitive data | Judges each situation and chooses Incognito chat / Connector boundaries accordingly |
8. Bridge to the Next Level
In L1 you learned to get a task done well in a single conversation. L2 takes the next step: turning Claude into a reusable workspace that remembers you — Projects, Memory, Skills for producing polished documents, and Connectors to your tools.