01 GAME LAB / TUTOR MODE

Make practice feel like play.

Build a small game for the exact thing your learner is finding hard—then notice what changes.

Start with a learner Why this works

For tutors, not game studios.Start with the lesson you already know needs work.

MAKE THE
NEXT TRY
EASIER

LESSON NOTE / 04 PLAYING

AMANI / READING

Short vowels

ROUND08

LISTEN FOR THE MIDDLE SOUND

What do you hear in ship?

Pick an answer to continue.

WHEN THEY MISS

Try one more word at the same level.

the game is small
so the noticing is clear

1 learnerin the brief

1 stubborn ideato practise

3 minutesto make a start

0 labelsattached to the child

THE FIRST EDUCATIONAL GAME / BEGINNER GUIDE

From a business idea to playable practice.

You do not need to become a game studio. You need one learner, one useful outcome, and a tiny first round you can watch someone play.

01 learner01 target01 round03 minutes

THE RECOMMENDED FIRST STACK

HTML / CSS / JavaScript + Google AI Studio + Codex + GitHub + Vercel.

Use the lightest stack that can make the learning moment playable. Add Cloud Code, Cloud Run, or Firebase only when you truly need a server, sign-in, saved progress, or live data.

THE TOOLKIT

Give every platform one clear job.

Most first games are easier to build when you separate thinking, making, history, and hosting.

01 / THINKSTART HERE

Google AI Studio

Explore an idea, test prompt wording, and turn a good prompt into starter code or a small prototype.

JOB: Shape the game brief and prompt.
USE WHEN: You are still deciding what the learner should do.

Read the official guide ↗

02 / MAKEBUILDER

Codex

Give the coding agent a repository and a precise task. It can inspect, edit, test, explain, and repair the smallest useful slice.

JOB: Turn the brief into working files.
USE WHEN: You have a clear first round to implement.

Learn Codex ↗

03 / EXTENDOPTIONAL

Cloud Code + Gemini Code Assist

Work inside VS Code, JetBrains, or Cloud Shell when your game needs Google Cloud services, debugging, or a backend.

JOB: Create, debug, and deploy cloud services.
USE WHEN: A static game is no longer enough.

Explore Cloud Code ↗

04 / REMEMBERALWAYS

GitHub

Keep the source, versions, notes, and collaboration in one repository so you can return to a working version.

JOB: Save history and share the project.
USE WHEN: Before the first meaningful change.

Read about repositories ↗

05 / SHIPPUBLISH

Vercel

Connect the repository, get preview deployments for changes, and attach a custom domain when the game is ready.

JOB: Put the game in front of a learner.
USE WHEN: You want a shareable link without server setup.

Read Vercel for GitHub ↗

06 / SAVELATER

Firebase

Add sign-in, saved progress, or a local emulator when a real learner account or data flow is genuinely needed.

JOB: Handle identity and persistent data.
USE WHEN: After the core learning loop works.

Start with Firebase Auth ↗

THE WORKFLOW

Build one round before a whole curriculum.

01
Observe
Write the exact moment that gets stuck: “mixes up short vowels,” “loses the sequence,” or “stops after a miss.”
02
Define the target
Choose one observable behavior. For example: “After three minutes, the learner identifies the middle sound in four out of five CVC words.”
03
Choose the mechanic
Match the action to the skill: sort, match, predict, route, build, or choose. Keep the rule explainable in one sentence.
04
Prompt the agent
Give the AI the learner, target, mechanic, constraints, and acceptance checks. Ask it to build the smallest vertical slice first.
05
Play before polishing
Test the first round yourself, then with the learner. Watch for hesitation, guessing, recovery, and whether feedback helps.
06
Publish and improve
Commit a working version, share the preview, change one thing, and compare what the learner does next time.

COPYABLE PROMPTS

Give the agent enough to make a good first move.

Paste one prompt, then replace the bracketed details. Review every generated answer and every generated code change.

01 / BRIEF

You are an educational game designer helping a business student tutor.
Learner: [age, subject, current confidence]
Learning target: [one observable behavior]
Game time: [2–5 minutes]
Mechanic: [sort / match / predict / route]

Design one playable round. Return:
1. the rule in one sentence;
2. the learner action;
3. three examples;
4. kind feedback for a miss;
5. how difficulty changes after two confident answers;
6. a simple success check.
Do not diagnose the learner or collect personal data.

02 / BUILD

You are a senior front-end engineer working in a small static HTML/CSS/JavaScript repository.

Build the smallest playable vertical slice for this brief:
[PASTE THE GAME BRIEF]

Constraints:
- no framework or backend for version one;
- responsive on a phone and laptop;
- keyboard accessible;
- one round, instant feedback, and a reset button;
- keep the visual design calm and readable;
- do not invent learner data.

Inspect the existing files first. Then implement only what is needed, explain the files changed, and list the checks I should run.

03 / AI STUDIO

Help me prototype an educational game prompt for Google AI Studio.

Subject: business basics
Learner: beginner student
Target: distinguish revenue, cost, profit, and break-even
Interaction: choose one answer in a short scenario
Tone: encouraging, precise, never childish

Return a structured game brief with:
- learning objective;
- one 3-minute round;
- question and choices;
- correct explanation;
- misconception-aware hint;
- a JSON example for one round.
Keep the output small enough to review by hand.

04 / CODEX

Inspect this repository before changing anything.

Task: implement [ONE SMALL CHANGE] for the educational game.
Acceptance criteria:
- [observable behavior 1]
- [observable behavior 2]
- works at 360px wide and desktop width;
- keyboard focus is visible;
- no console errors.

Show me the files you plan to change, make the smallest safe patch, run the relevant checks, and report anything you could not verify. Do not refactor unrelated code.

05 / CLOUD

The static educational game now needs [saved progress / a small API / sign-in].

Propose the smallest Google Cloud path using Cloud Code and Gemini Code Assist.
Compare:
1. keeping the game static;
2. adding a small Cloud Run service;
3. using Firebase Auth and a database.

For the recommendation, name the files, environment variables, data to store, local test plan, deployment steps, and privacy risks. Never put secrets in client-side code. Do not add cloud services until the learning loop is already tested.

06 / QA

Act as a careful QA reviewer for this educational game.

Test and report:
- the learning rule is clear before play;
- correct and incorrect choices give useful feedback;
- a learner can recover after a miss;
- keyboard-only play works;
- focus and contrast are visible;
- the layout works at phone width;
- refresh does not create false progress;
- no personal learner data is exposed;
- there are no console errors.

Return a short table of pass, fail, evidence, and the smallest fix.

SHIP CHECK

Ready to let a learner try?

Tick the basics before you send the link. A small, clear game is more useful than a feature-heavy one nobody can finish.

One learner and one observable target are named. The first round can be explained in one sentence. Every choice gives feedback that helps the next try. The game works with a keyboard and on a phone. No diagnosis or unnecessary personal data is collected. You played it, reviewed the AI output, and saved a version.

0 of 6 complete

Human review stays in the loop. Game performance is a clue for your next teaching move—not a label, diagnosis, or verdict about a child.

WHY GAME LAB EXISTS

The hard part is already in your notes.

You have seen the pause, the guess, the “I don’t know” before the question is finished.

A game gives that moment somewhere to go. Turn the observation into one clear choice, a little feedback, and a reason to try once more.

KEEP THE LEARNERIN THE BRIEF.

THE TUTOR LOOP

Make one small thing easier to try.

Good practice does not need a bigger scoreboard. It needs a clear next move.

Notice

Name the exact thing that is hard: blending sounds, holding a sequence, spotting a pattern, staying with a task.

START WITH THE OBSERVATION

Make

Give the skill a rule, a moment of choice, instant feedback, and a reason to try the next round.

MAKE PRACTICE PLAYABLE

Notice again

Watch where the learner speeds up, stalls, guesses, or recovers. That pattern is your next lesson plan.

LET THE DATA STAY HUMAN

STARTER GAMES

Pick the difficult bit.

Start from the skill, then let the theme do the motivating. These are practical starting points—not diagnoses.

01: Reading
02: Maths
03: Focus

STARTER / 01 READY TO TRY

Word Hopper

A quick-fire sound-matching run for short vowels, blending, and confidence in the first read.

PLAYChoose · hear · repeat

KEEP GOINGUnlock the next word set

FIRST PROMPTREADY

“Build a 3-minute game for a learner practising short vowels. Keep the screen calm, use one word at a time, give audio-friendly feedback, and adjust the next word after two misses.”

WHAT TO WATCH

Good data leaves room for judgement.

Use performance as a clue, not a verdict. The point is to make the next useful move easier to see.

PRIVATE TUTOR NOTE / EXAMPLEAMANI — CVC WORDS

AFTER 3 SESSIONS

Word recognition82%fewer guesses

Response time3.1smore settled

Recovery after miss6 / 8try a shorter round

PLAY SESSION SIGNAL01 — 04

THE NEXT LESSON

Bring one difficult thing.

One learner. One stubborn concept. One small game. That is enough to start.

Choose a starter game