Claude 101

A free, beginner-friendly introduction to Claude AI - prompts, projects, artifacts and Claude Code for software testers and engineers

Course Summary

Claude 101 is a completely free starter course that takes you from zero to confident with Anthropic's Claude. You will learn what Claude is, how the chat, Projects, Artifacts and Claude Code surfaces work, how to write prompts that actually produce usable output, and how to apply Claude to everyday software testing and engineering work. Finish the two lessons and the short quiz to unlock your certificate of completion.

Course Curriculum

Why Claude 101 exists

Everyone tells you to "use AI". Almost nobody shows you how.

You have probably opened Claude, typed a question, got a reasonable-looking answer, and quietly gone back to doing the work yourself. That is not a failure of the tool - it is what happens when a genuinely capable system is used through a one-line prompt with none of the context it needs. Claude 101 fixes that in about an hour.

This is a completely free, beginner-first course. No prior AI experience, no coding, no credit card. You will learn what Claude actually is, how its four surfaces work, how to write prompts that produce usable output on the first attempt, and how to apply all of it to everyday software testing and engineering work. Finish the five lessons and you get a certificate of completion.

"I asked it and the answer was generic"
Because the model could not possibly guess your stack, your conventions or your definition of done. You will learn exactly what to put in and why it changes everything.
"I don't know which model or plan to use"
You will learn the trade-off between speed and reasoning depth, what the free tier really gives you, and when paying is actually justified.
"I keep re-explaining my project every time"
Projects and custom instructions solve this permanently. Ten minutes of setup replaces hours of repetition every week.
"It made something up and I nearly shipped it"
You will learn Claude's failure profile, the six-step verification routine, and how to make it safe for the model to say it does not know.
"I don't know what is actually worth delegating"
You get eight concrete jobs that work, an honest list of what does not, and a simple rule based on verification cost.
"My team wants a policy and I have no idea"
You get a paste-ready safe-usage rule covering what may and may not go into a consumer-tier AI tool.
What you will be able to do

By the end of one hour, you will work with Claude like someone who has used it for a year

These are the concrete, checkable outcomes of the course. Each one maps to a section of the written notes and appears in the self-assessment in the final lesson.

1
Explain what Claude is doing
Describe in one sentence what a language model is actually doing when it answers - and why that explanation changes how you prompt.
2
Choose the right surface
Know when to use a throwaway chat, a persistent Project, an Artifact, or Claude Code - instead of doing everything in one chat box.
3
Write a C.R.I.S.P. prompt
Turn a vague one-line request into a structured prompt that gets a usable answer on the first attempt, every time.
4
Ground answers in real evidence
Attach specifications, logs and screenshots and get answers with verbatim citations instead of confident guesses.
5
Build a reusable Project
Write custom instructions carrying your stack, conventions and output rules so you never re-explain your context again.
6
Triage a real failure
Run the three-hypothesis triage template on a genuinely flaky test and know the exact check to confirm each cause.
7
Generate awkward test data
Produce synthetic datasets that include the unicode, timezone, leap-year and rounding cases you would have forgotten.
8
Spot a hallucination
Recognise Claude's failure profile and run a six-step verification routine on anything that matters.
9
Decide what not to delegate
Apply the verification-cost rule so you stop wasting time checking work you should have done yourself.
How the course works

Five short lessons, one hour, then a seven-day practice plan

Claude 101 is deliberately small. The value is not in the hours - it is in the structure, the templates, and the habit you build in the week that follows.

1
Read the written notes
Three parts covering the mental model, the four surfaces, C.R.I.S.P. prompting, grounding, the eight applied jobs, and the limits.
2
Watch the two videos
Everything demonstrated on screen against a real application and a real failing test - including the part where the first answer is wrong.
3
Take the prompt library
24 copy-paste templates across six categories, each written in the C.R.I.S.P. structure and ready to be edited for your project.
4
Run the 7-day plan
Fifteen minutes a day on real work. Setup, requirements, test design, triage, review, communication, and an honest reflection.
5
Claim your certificate
Complete all five lessons, including both compulsory video lessons, and your certificate is issued automatically.
Curriculum in detail

Exactly what is inside each of the five lessons

Nothing is padded and nothing is hidden behind an upsell. This is the complete contents of the free course.

1

1. Start Here - Complete Written Guide to Claude 101

45 min read

The full written companion, delivered in three parts so it stays readable. Part 1 covers what Claude actually is, how to get set up safely, and the four surfaces. Part 2 is the prompting skill: the C.R.I.S.P. structure, few-shot examples, chain-of-thought, and working from real files instead of memory. Part 3 is application and judgement: eight jobs Claude does well, where it fails, and how to verify anything that matters.

  • ▸What an LLM is really doing, and why that changes your prompts
  • ▸Tokens, context windows and why very long chats degrade
  • ▸Free tier versus paid, and a paste-ready data-handling rule
  • ▸Chat vs Projects vs Artifacts vs Claude Code
  • ▸Writing Project instructions that actually steer output
  • ▸The C.R.I.S.P. prompt structure with a full before-and-after
  • ▸Few-shot examples and step-by-step reasoning
  • ▸Grounding: attachments, citations and the NOT SPECIFIED escape hatch
  • ▸Map-reduce for very large documents
  • ▸Eight applied jobs from requirement analysis to onboarding docs
  • ▸Hallucinations, sycophancy, non-determinism and prompt injection
  • ▸A six-step verification routine and a full 24-term glossary
You build: A safe-usage policy for your team, a Project instruction block for your stack, and a verification routine you will actually follow.
2

2. Video 1 - Claude Fundamentals: Chat, Projects, Artifacts & Files

Video - compulsory

A screen-recorded walkthrough from a genuinely blank account. The interface, the model selector, the settings that matter, and then the centrepiece: creating a Project, writing real custom instructions, attaching knowledge, and asking the same question inside and outside the Project so the difference is visible rather than described.

  • ▸The interface: composer, sidebar and model selector
  • ▸Your first prompt, and why the answer is mediocre
  • ▸Choosing a model: speed versus reasoning depth
  • ▸Usage limits and data controls, explained honestly
  • ▸Creating a Project from scratch
  • ▸Custom instructions that steer output rather than describe a personality
  • ▸Project knowledge, and testing that it is actually being used
  • ▸Artifacts: generate, revise, version
  • ▸Where Claude Code fits and why it is out of scope here
You build: Your first real Project, with instructions and at least one reference document attached.
3

3. Video 2 - Claude for Testers: Prompting, Claude Code & Real Workflows

Video - compulsory

The applied video. The before-and-after prompt comparison, ambiguity hunting on a real requirements document, and a long live triage of a Playwright test that passes locally and fails forty percent of the time in CI - including running the confirming check for hypothesis one and finding it wrong.

  • ▸The bad prompt, and why its answer is unusable
  • ▸The same request rebuilt with all five C.R.I.S.P. elements
  • ▸Locking a format with a single few-shot example
  • ▸Ambiguity hunting on a real requirements document
  • ▸Test case generation from corrected requirements
  • ▸Synthetic test data, including the awkward rows
  • ▸Three-hypothesis triage on a genuinely flaky test
  • ▸Being wrong first time, and working down the ranked list
  • ▸Reviewing test code against written conventions
  • ▸A release summary written for a named audience
  • ▸Claude Code in a repository: a multi-file change
  • ▸What was NOT worth delegating, and why
You build: A triaged real failure of your own, and a requirement-analysis pass on one of your own tickets.
4

4. Prompt Library & Cheat Sheet - Ready-to-Use Claude Prompts for QA

Reference

Twenty-four copy-paste templates across six categories, every one written in the C.R.I.S.P. structure with angle-bracket placeholders for your own detail. This is the lesson people come back to weeks later.

  • ▸Requirements and test design: ambiguity hunter, test case designer, risk prioritiser, acceptance criteria writer
  • ▸Test data and environments: synthetic datasets, boundary values, fixtures, environment diff
  • ▸Debugging and flakiness: three-hypothesis triage, flaky autopsy, stack trace explainer, bug report writer
  • ▸Code review and migration: convention-based review, flakiness audit, migration with an example, page object extractor
  • ▸Documentation and reporting: release summary, runbook, onboarding doc, escalation message
  • ▸Learning and career: levelled explanation, quiz me, codebase tour, interview rehearsal
You build: Your own edited versions of at least five templates, saved somewhere you will find them again.
5

5. Wrap Up - Your 7-Day Claude Practice Plan & Certificate Checklist

15 min

A day-by-day plan that takes fifteen to twenty minutes a day on real work, a twenty-item self-assessment mapped back to the written notes, and the exact requirements for claiming your certificate.

  • ▸Day 1: set up a Project that carries your real context
  • ▸Day 2: run the ambiguity hunter on a live ticket
  • ▸Day 3: generate test cases and honestly count what you kept
  • ▸Day 4: triage a real failure and time it
  • ▸Day 5: convention-based review, and watching it fold when you push back
  • ▸Day 6: write a release summary for a named person and send it
  • ▸Day 7: deliberately make it hallucinate, then decide what you will not delegate
  • ▸A twenty-item self-assessment with the notes section for each gap
  • ▸Exactly what unlocks the certificate
You build: A seven-day log you can read back as evidence rather than a vague feeling.
What you will be working with

Tools and concepts covered

Everything here is available on the free tier of claude.ai. Nothing in this course requires a paid plan or an API key.

Tool or conceptHow it is used in this course
claude.ai chatThe disposable surface for single questions. Where you start and where most people stop.
ProjectsPersistent workspaces with custom instructions and knowledge files. The highest-leverage feature in the whole product.
ArtifactsVersioned, editable output for documents and code you intend to keep and revise.
File attachmentsSpecifications, logs, spreadsheets and code files used to ground answers in evidence.
Vision / screenshotsTurning UI screenshots and whiteboard sketches into bug reports and written specs.
Claude CodeIntroduced only - the terminal-based agentic tool that works inside a repository. Shown so you know where this goes next.
C.R.I.S.P.Context, Role, Instruction, Shape, Proof. The prompt structure used in every template in the course.
Few-shot promptingOne or two worked examples to lock an output format far more reliably than description can.
Chain-of-thoughtRequesting reasoning before conclusion on anything involving logic, dates, counting or state.
Grounding and citationsVerbatim source quotes plus a NOT SPECIFIED escape hatch to stop the model filling gaps.
Playwright and TypeScriptUsed only as the example stack in the applied examples. You do not need to know it to follow along.
Verification routineA six-step check applied to any output that has real-world consequences.
Before you start

What you need - which is almost nothing

This is a genuine beginner course. If you can open a browser and you have worked on software in some capacity, you have everything you need.

Required
  • ✓A free claude.ai account - the free tier is enough for the entire course
  • ✓A browser and about one hour
  • ✓Some familiarity with how software gets built and tested, at any level
  • ✓A real piece of work - a ticket, a spec, a failing test - to practise on
Helpful but not required
  • +Any programming language at a basic level, for the code examples
  • +Some exposure to test automation, since the applied examples use Playwright
  • +Permission from your employer to use an AI tool on work material
  • +A team you can share the Project instruction block with afterwards
No credit card, no trial, no upsell at the end. Claude 101 is free permanently, and the certificate is free too. The only thing you have to spend is an hour of attention.
Fit check

Who this course is for - and who should skip it

A free course still costs you an hour, so it is worth being honest about who benefits.

This is for you if
  • ✓You have never used Claude, or you have used it casually and suspect you are getting ten percent of the value
  • ✓You are a tester, QA engineer, SDET or developer who wants concrete workflows rather than general AI theory
  • ✓You keep re-explaining your project context in every conversation and want that to stop
  • ✓You have been burned by a confident wrong answer and want to know how to catch the next one
  • ✓You need to explain to your team or your manager what is safe to put into an AI tool
  • ✓You want a small set of templates you will actually reuse, not a 12-hour course you will never finish
  • ✓You are about to take a deeper AI or automation course and want the fundamentals solid first
This is probably not for you if
  • ✗You already write structured prompts, use Projects daily and verify output systematically
  • ✗You want a deep dive on the API, function calling, embeddings or fine-tuning - start with LLM Fundamentals for QA instead
  • ✗You are looking for Claude Code in depth - it is introduced here, not taught
  • ✗You want a tool that replaces judgement rather than compressing the mechanical middle of your job
What you take away

What you actually keep after the hour

Course completion is worthless on its own. These are the artefacts you walk away with and use next week.

A Project instruction block
Role, stack, conventions and output rules for your product area - written once, inherited by every conversation.
24 prompt templates
Six categories, all in the C.R.I.S.P. structure, edited by you for your own project.
A safe-usage policy
A paste-ready rule for what may and may not go into a consumer-tier AI tool, ready for your team wiki.
A triage playbook
The three-hypothesis template plus the discipline of running the confirming check before acting.
A verification routine
Six checks to run on any output that has real-world consequences, and a rule for what not to delegate.
A 7-day practice log
Evidence of where this actually saved you time, in your own work, rather than someone else's demo.
A full glossary
Twenty-four terms defined plainly, so the next AI course you take does not lose you in the first ten minutes.
A certificate
Issued automatically once all five lessons, including both video lessons, are complete.
Before and after

The same hour of work, before and after Claude 101

This is the honest difference the course makes. Not magic - just structure applied consistently.

TodayAfter this course
You type "write test cases for login" and get something genericYou supply context, role, format and acceptance criteria and get a table you can mostly use as-is
You re-explain your stack and conventions in every conversationYour Project instructions carry them automatically into every conversation
You paste a stack trace and get a plausible-sounding guessYou get three ranked hypotheses, each with the exact check that confirms or eliminates it
You accept a confident answer because it sounds rightYou run a six-step verification routine on anything that matters
You ask "are you sure?" and treat agreement as confirmationYou ask it to argue the opposite case, or re-ask in a fresh conversation
You avoid attaching documents because it feels slowYou ground every serious question in a real source and demand verbatim citations
You have no idea what is safe to paste at workYou have a written policy you can defend to your security team
You delegate whatever feels tedious and then re-do itYou delegate based on verification cost, and keep the judgement work yourself
Your instructor

Pramod Dutta

Pramod Dutta is the founder of TheTestingAcademy and has spent his career in software testing and test automation, working with teams on Selenium, Playwright, API testing, CI/CD pipelines and, more recently, AI-assisted testing workflows.

He has taught tens of thousands of testers through TheTestingAcademy courses and his YouTube channel, and the teaching style throughout is deliberately practical: real tickets, real failing tests, real conventions, and an honest account of the parts where the tool gets it wrong.

Claude 101 exists because the same three questions kept coming up from students starting the deeper AI courses - what is this thing actually doing, how do I prompt it properly, and what is safe to paste at work. This course answers all three in an hour, for free.

50K+
Students taught
10+
Years in testing
8
Courses on this platform
Free
Claude 101, permanently
Questions people ask

Frequently asked questions

If your question is not here, ask on the community and it will usually be answered the same day.

Is this really free, or is there a catch at the end?
It is genuinely free, permanently, and so is the certificate. There is no card required and no upsell inside the lessons. There is a list of deeper paid courses at the end, which you are free to ignore.
Do I need a paid Claude subscription?
No. Everything in this course works on the free tier of claude.ai. The course explains what the paid tiers add so you can decide later, but you will not hit a wall while learning.
Do I need to know how to code?
No. The applied examples use Playwright and TypeScript, but they are explained rather than assumed, and none of the exercises require you to write code yourself.
I am not a tester. Is this still useful?
Yes. Sections 1 to 5 are entirely general - what Claude is, how to set it up, the four surfaces, prompting and grounding. Only section 6 and the prompt library are testing-specific, and the structure transfers to any technical role.
How long does it actually take?
About an hour to read the notes and watch both videos. The seven-day practice plan then takes fifteen to twenty minutes a day, on work you were going to do anyway.
How do I get the certificate?
Complete all five lessons. Both video lessons are marked compulsory, so they must be watched through rather than opened. Once everything is complete the certificate is generated automatically and appears in your account as a downloadable PDF.
Is it safe to put my work code into Claude?
That depends entirely on your plan and your employer's policy, and the course covers this properly. You get a paste-ready rule separating what is fine from what needs an enterprise agreement, plus guidance on reading the data controls page.
Will this be out of date in six months?
The model names will move; the skills will not. C.R.I.S.P., grounding, verification and the delegation rule apply to every model from every vendor. The course is reviewed and refreshed as the product changes.
What should I take after this?
LLM Fundamentals for QA for the concepts underneath, then Playwright MCP Masterclass and the Playwright AI Agents track for the applied automation work.
Can my whole team take it?
Yes, and it works better that way. The Project instruction block and the safe-usage policy are both team artefacts - agreeing them together is most of the value.
Start now

One hour, no cost, and you stop guessing

Enrol free, read the notes, watch the two videos and run the seven-day plan on your own work. If it does not change how you work within a week, you have lost an hour.

  • ✓Free forever - no card, no trial, no upsell
  • ✓5 lessons, 3-part written guide, 2 videos
  • ✓24 copy-paste prompt templates
  • ✓Certificate of completion included
  • ✓Works entirely on the free Claude tier

Pramod Dutta

Founder of The Testing Academy, a YouTube Channel with 95K subscribers where Pramod teaches about Software testing & Test Automation. With overall 10+ years of experience in Software Testing & Test Automation, he has mentored 10,000+ students in Software Testing, API Testing, Test Automation.

Pramod Dutta is working as SDET Manager at Tekion | Ex BrowserStack Employee & BrowserStack Champion & Certified Scrum Master. Pramod has a vast range of experience handling from Manual Testing, Mobile & Web Automation, Desktop & cloud services like AWS, GCP extra.

By joining this Automation Testing Course , you’ll have the opportunity to take control of your life, work in an exciting industry with infinite possibilities and live the life you want.

John Smith

Developer

Highly Recommended Course. Easy to Understand, Informative, Very Well Organized. The Course is Full of Practical and Valuable for Anyone who wants to Enhance their Skills. Really Enjoyed it. Thank you!!

Course Pricing

Free Access

Free

  • Free forever. Full access to all five lessons, the complete written notes, both video lessons and the prompt library - plus your c

    • 5 lessons covering Claude from zero to confident
    • 2 video lessons: Claude Fundamentals and Claude for Testers
    • Complete written notes in three parts, plus a full glossary
    • 24 copy-paste prompt templates for real QA work
    • Certificate of completion - no card, no upsell
Enrol Free

© 2026 TheTestingAcademy