claude-mastery-course

# 📚 Module 09 — Resources, Cheat Sheets & Glossary [![Module](https://img.shields.io/badge/09-Resources-D97757?style=for-the-badge&labelColor=2a1f1a)](#) [![Reference](https://img.shields.io/badge/📖_Reference-FFD23F?style=for-the-badge&labelColor=2a1f1a)](#) [![Level](https://img.shields.io/badge/✨_All_levels-6BCF7F?style=for-the-badge&labelColor=2a1f1a)](#) ***The closing reference module — bookmark this one.***

[!TIP] 🎨 Looking for visuals? → Visual Reference — every diagram from the course in one page

🖥️ Want it interactive? → Open ../hub.html in your browser

9.1 Official Anthropic Resources

What	URL
Claude.ai	https://claude.ai
Claude API docs	https://docs.claude.com/en/api/overview
Models overview	https://docs.claude.com/en/docs/about-claude/models/overview
Pricing	https://www.anthropic.com/pricing
Console & API keys	https://console.anthropic.com
Workbench (prompt testing)	https://console.anthropic.com/workbench
Claude Code docs	https://docs.claude.com/en/docs/claude-code/overview
Claude Code on npm	https://www.npmjs.com/package/@anthropic-ai/claude-code
Anthropic Cookbook (notebooks)	https://github.com/anthropics/anthropic-cookbook
Help Center	https://support.claude.com
Anthropic news / launches	https://www.anthropic.com/news
MCP (Model Context Protocol)	https://modelcontextprotocol.io
Status page	https://status.anthropic.com

9.2 The Prompt Engineering Cheat Sheet

🎭 ROLE         "You are a [specific] who has done [thing] for [time]."
📋 CONTEXT      "I'm building/writing/deciding about [situation]."
🎯 TASK         "Do exactly: [verb] + [object] + [success criterion]."
📐 CONSTRAINTS  "Length / tone / what to avoid / what must include."
📦 FORMAT       "Output as: markdown table / JSON / 5 bullets / etc."
🌟 EXAMPLE      "Match the tone of: [paste]."

Power moves

XML tags for structured input: <context>...</context>, <draft>...</draft>
Step-by-step thinking for hard problems
Few-shot examples (3 input/output pairs) for classification
“Ask me 3 questions first” for ambiguous tasks
“What would you change to make this 50% better?” for iteration

9.3 The API Cheat Sheet

from anthropic import Anthropic

client = Anthropic()

# --- BASIC CALL ---
resp = client.messages.create(
    model="claude-sonnet-4-6",   # or claude-opus-4-7, claude-haiku-4-5
    max_tokens=1024,
    system="You are a concise assistant.",
    messages=[{"role": "user", "content": "Hello"}],
    temperature=0.7,             # 0 = deterministic, 1 = creative
)
print(resp.content[0].text)

# --- STREAMING ---
with client.messages.stream(model=..., max_tokens=..., messages=...) as s:
    for chunk in s.text_stream:
        print(chunk, end="")

# --- VISION ---
{"type": "image", "source": {"type": "base64", "media_type": "image/png", "data": b64}}

# --- PDF ---
{"type": "document", "source": {"type": "base64", "media_type": "application/pdf", "data": b64}}

# --- TOOL USE ---
client.messages.create(..., tools=[{"name": "...", "description": "...", "input_schema": {...}}])
# stop_reason will be "tool_use"; loop until "end_turn"

# --- PROMPT CACHING ---
system=[
    {"type": "text", "text": "...", "cache_control": {"type": "ephemeral"}},
]

9.4 Choosing a Model — Quick Decision Tree

Is the task simple/repetitive? ──Yes──► Haiku
       │
       No
       │
Is it a hard reasoning/coding task ──Yes──► Opus (or Opus + extended thinking)
or a long-horizon agent?
       │
       No
       │
       └──► Sonnet (the default for most production work)

9.5 Glossary

Token — A chunk of text the model processes (roughly ¾ of a word).
Context window — How much text the model can hold at once. Modern Claude: up to 1M tokens.
System prompt — A persistent instruction that sets Claude’s role/rules.
Temperature — Randomness control. Low = consistent, high = creative.
Streaming — Returning the response token-by-token as it’s generated.
Tool use / function calling — Claude calling your code via a defined interface.
Agent — Claude in a loop, using tools to make progress on a goal.
MCP (Model Context Protocol) — Open standard for connecting AI to tools/data.
Artifact — A live document, code file, or app rendered in Claude.ai.
Project — A persistent workspace with shared instructions and files.
Prompt caching — Reusing pre-computed input to cut cost & latency.
Extended thinking — A mode where Claude reasons explicitly before answering.
Hallucination — When the model makes something up that sounds plausible but isn’t true.
RAG — Retrieval-Augmented Generation. Fetch relevant docs, then answer.
Constitutional AI — Anthropic’s training method using AI-generated feedback against a set of principles.
RLHF — Reinforcement Learning from Human Feedback.

9.6 FAQ

Q: Will Claude train on my data? A: For Claude.ai consumer plans, you control this in Settings — opt out of training is supported. For API and enterprise plans, your data is not used to train models by default. Always read the current terms.

Q: Is the API the same as Claude.ai? A: No. The API is raw access to the model. Claude.ai is a polished product built on top of the API with extra features (Projects, Artifacts, search, memory). Pricing is also separate.

Q: How do I report a bad response? A: Thumbs-down in Claude.ai sends feedback to Anthropic. For specific safety issues, see anthropic.com/safety.

Q: How do I keep up with new features? A: anthropic.com/news and the release notes.

Q: Can I run Claude locally / offline? A: No. Claude is closed-source and runs on Anthropic’s infrastructure. For self-hosting, look at open models like Llama, Mistral, or Qwen.

Q: What’s the difference between Claude and ChatGPT? A: Different companies (Anthropic vs. OpenAI), different training approaches (Constitutional AI vs. RLHF heavier), different strengths (Claude often praised for writing/coding/long context; GPT for ecosystem integration). Try both — they’re complementary.

9.7 Recommended Learning Path After This Course

Build something that solves a real problem for you — even a tiny one. The biggest gains come from real use, not more reading.
Read the Anthropic Cookbook — concrete examples for tool use, RAG, evals.
Join the Anthropic Discord — real engineers, real questions, no marketing fluff.
Follow the release notes — this space moves fast.
Keep a personal prompt library. Your best prompts are an asset. Save them.

9.8 Contributing & Feedback

Found a bug, have a better example, or want to add a project? PRs welcome — see CONTRIBUTING.md.

🎉 You finished the course

If you’ve actually built the projects in Module 08, you’re now in the top few percent of Claude users. Most people stop at “type a question, copy the answer.” You’ve moved from a user to a builder.

Now go ship something.

⭐ If this course helped, please star the repo so other people can find it.

← Previous	🏠 Home	Next →
Module 08 — Real-World Projects	Course README	End of course

This site is open source. Improve this page.