Anthropic interactive prompt engineering tutorial
Anthropic interactive prompt engineering tutorial
Well structured with exercises; you need to be comfortable with notebooks to learn from it.
Notebook-style lessons from beginner to advanced, with hands-on exercises after each chapter.
Overview
This is a prompt-writing course published by Anthropic on GitHub, with 9 chapters with exercises and an appendix, going from basic prompt structure to building complex prompts for industry scenarios. The lessons are Jupyter Notebooks, with an experimentation area at the end of each lesson and reference answers.
The course was written for Claude 3 models (Claude 3 Haiku by default). Anthropic's documentation notes that where the course differs from the current Prompting best practices guide, you should follow the newer guide. Besides the notebook version, there is a Google Sheets version that the README describes as easier to use.
For businesses, this is a systematic way to learn: understanding why prompts give wrong results, how to separate data from instructions and how to reduce made-up information, before putting AI into real workflows.
Highlights
Three-level path
Beginner (basic structure, clarity, assigning roles), intermediate (separating data, output formatting, step-by-step thinking, examples) and advanced (avoiding hallucinations, complex prompts).
Hands-on exercises
Each chapter has exercises and an Example Playground area where you edit prompts yourself and see results right away.
Industry scenarios
Chapter 9 has chatbot and legal-service examples plus exercises on finance and coding.
Advanced appendix
Introduces prompt chaining, tool use and retrieval search.
Multiple versions
Folders for the Anthropic API (Anthropic 1P), Amazon Bedrock and a Google Sheets version.
Answers included
A static answer key is available so you can study without an API key.
What it's for
Train employees to use AI
The company assigns a core group to study the chapters in order, then share what they learned with the whole company.
Prepare before automating
Technical teams learn to separate data and format output before putting Claude into order or email processing flows.
Reduce wrong answers
Apply the Avoiding Hallucinations chapter when using AI to answer questions about policies and products.
Build an advisory chatbot
Use chapter 9 to build a multi-part prompt for a customer-support chatbot.
How to use it
- 1
Choose a version
Non-programmers should use the Google Sheets version; those who know Python use the Jupyter Notebook version in the Anthropic 1P folder.
- 2
Download the source code
Clone the repository to your computer following the instructions in the 00_Tutorial_How-To notebook.
- 3
Install the library
In the first notebook, run the cell that installs the Anthropic SDK.
$!pip install anthropic - 4
Set the API key and model
Replace the API_KEY value with your key from the Anthropic Console; the notebook presets MODEL_NAME to Claude 3 Haiku.
- 5
Study the chapters in order
Start with 01 Basic Prompt Structure, run each cell from top to bottom and do the exercise at the end of each lesson.
Video guides
Anthropic just dropped a Free full Prompt Engineering course on this Github Repo
Prajwal Tomar
Prompting 101 Code w/ Claude
Anthropic
Real-world example
Hypothetical example: a home appliance distributor wants to use AI to answer dealer emails. Before starting, the IT person spends a few evenings on this course, focusing on the chapters on separating data from instructions and avoiding hallucinations. He then writes a prompt that puts the pricing policy in a separate block, requires the AI to answer only from that document and say clearly when it can't find the information, and has the sales team test it on old emails before real use.
Tips & notes
Study in the right order
Later chapters reuse knowledge from earlier ones, so don't skip around.
Check against newer docs
The course was written for Claude 3; for current models, also see Prompting best practices where they differ.
Protect your API key
Don't share a notebook containing a real API key in public places.
Use the Playground
Rewrite prompts using your own business's examples to remember them longer.
FAQ
Do I need an API key?
The notebook version needs an Anthropic API key to run; if you don't have one, you can view the static answer key or use the Google Sheets version.
Does it cost money?
The course materials are public on GitHub; Claude API calls are billed according to Anthropic's pricing. See the terms on the official site.
Can I learn without knowing Python?
The README suggests the Google Sheets version is easier, suitable for non-programmers.
Is there a Vietnamese version?
The content is in English.
Useful links
Want to apply this in your business?
2dm looks at your workflows, picks the right tools and sets them up with your team. The first call is free.
Same topic
2dm Prompt Library
Prompts for marketing, sales, operations and learning. Fill in a few fields, then copy or open directly in ChatGPT or Claude.
Tailor-made sales landing page prompt
A prompt for Claude Code or a coding agent: build a sales landing page with its own identity for each product, honest about figures, with a lead form, stopping at a draft.
Claude prompting guide
4.5
Anthropic's official guide: how to write clear requests, give examples, break down tasks and let Claude think step by step.