DEV Community: Yousun Lotif (Bappy Khan)

Turing’s Solstice by Yousun — An AI-Powered Puzzle Game Where Every Hint Costs Light

Yousun Lotif (Bappy Khan) — Sat, 20 Jun 2026 09:21:39 +0000

This is a submission for the June Solstice Game Jam

Turing’s Solstice by Yousun

Decode the Last Light.

What if asking for help could bring you closer to losing?

Turing’s Solstice is a cinematic browser-based puzzle game inspired by the June Solstice, Alan Turing, code-breaking, and the fragile balance between daylight and darkness.

The player must solve six increasingly complex puzzles before the Daylight Meter reaches zero. Correct answers restore light, wrong answers strengthen the darkness, and every Gemini AI hint costs part of the player’s remaining daylight.

The result is a puzzle journey where logic is not the only challenge—resource management, risk, and timing matter too.

What I Built

I built a six-level code-breaking adventure where players restore a damaged solstice machine by recovering hidden fragments.

The journey includes:

Solar Cipher — decode a Caesar-shifted message
Binary Stream — translate binary into language
Algorithm Fragment — recognize a mathematical pattern
Mirror Solstice Cipher — reverse a shadow message
Turing Decision Gate — solve a light-and-dark logic sequence
Final Solstice Key — combine the recovered knowledge into the final machine key

Every solved level unlocks:

A recovered machine fragment
A Turing-inspired note
A daylight reward
A cinematic transition into the next challenge

The intended experience is short, focused, replayable, and atmospheric—something that feels more like entering a mysterious decoding machine than completing a normal web form.

How It Connects to the June Solstice

The June Solstice represents the longest day and the relationship between light and darkness.

I turned that theme into the core gameplay system.

The Daylight Meter acts as both energy and survival time:

Correct answer: +10% daylight
Wrong answer: -12% daylight
Gemini hint: -8% daylight
Daylight reaches zero: the machine loses balance

This means the solstice theme is not only visual decoration. It directly controls how the player thinks, takes risks, and progresses.

The golden side of the interface represents daylight, while the blue side represents night, computation, and uncertainty.

Video Demo

Direct video link:

https://youtu.be/QtAMMGERbeQ

The demo shows:

The home and project story pages
The cinematic level-intro system
Puzzle solving
Daylight loss and recovery
Gemini-powered hints
Adaptive Oracle modes
Fragment recovery animations
Sound and visual feedback
The final restoration sequence

Live Demo

Play Turing’s Solstice by Yousun here:

https://turing-s-solstice-by-yousun.vercel.app/

Code

GitHub repository:

https://github.com/yousunlotif-bappy/Turing-s-Solstice-by-Yousun

How I Built It

The project was built with:

Next.js
React
TypeScript
Google Gemini API
CSS animations
Web Audio API
Vercel

I used the Next.js App Router to separate the experience into:

Landing page
Game page
Project story page
Server-side Gemini hint API route

The puzzle content is stored as structured TypeScript data. Each level contains its title, puzzle type, objective, answer, recovered fragment, success note, and Turing-inspired archive note.

This structure made it easy to expand the game from four levels to six without rewriting the main game engine.

Core Game State

React state controls the main gameplay systems:

Current level
Daylight percentage
Player answer
Wrong-attempt count
Number of hints used
Current Gemini hint
Adaptive hint mode
Recovered fragment
Unlocked Turing notes
Reward animation
Level-intro overlay
Win and lose states

Answers are normalized before comparison, so reasonable variations are accepted.

For example, the final key can be entered with different capitalization, spacing, or separators without unfairly rejecting the player.

Google Gemini as the Solstice Oracle

Gemini is not used as a generic chatbot.

Inside the game, Gemini becomes the Solstice Oracle—an adaptive in-world guide that helps the player without directly revealing puzzle answers.

Every hint request sends gameplay context to the server:

Current level
Puzzle type
Puzzle text
Remaining daylight
Wrong-attempt count
Number of hints already used

The Oracle then responds in one of three modes.

Cryptic Mode

Used when the player has high daylight and few mistakes.

The hint remains mysterious and indirect.

Balanced Mode

Used when the player has made a mistake or daylight has started falling.

The hint becomes more practical while still requiring thought.

Urgent Mode

Used when daylight is low or the player has made multiple mistakes.

The hint becomes clearer, but the Oracle is still instructed never to reveal the final answer.

This creates an AI system that responds to the player’s situation instead of returning the same type of hint every time.

Most importantly, every request costs daylight.

The player must decide:

Should I preserve the remaining light and solve the puzzle myself, or sacrifice daylight for guidance?

That decision makes Gemini part of the game’s strategy, tension, and economy.

Reliable Fallback System

The game also includes level-specific fallback hints.

If the Gemini API key is unavailable, the request fails, or the model returns no usable text, the game still provides a safe hint instead of breaking.

This keeps the project playable in local, preview, and production environments.

The Gemini API key remains server-side through environment variables and is never exposed in client-side code.

Fragment Recovery System

Correct answers do more than move the player to the next level.

Each success triggers a reward moment:

FRAGMENT RECOVERED

DAYLIGHT RESTORED +10%

TURING NOTE UNLOCKED

MACHINE STABILITY IMPROVED

This gives every solved puzzle a visible narrative purpose.

The player is not simply answering questions—they are rebuilding the solstice machine one fragment at a time.

Alan Turing Inspiration

Alan Turing’s legacy shaped the entire puzzle journey.

The game explores:

Caesar ciphers
Binary communication
Pattern recognition
Algorithms
Logical state decisions
Code-breaking
Machine-like reasoning

Each completed puzzle unlocks a short Turing Note connected to computation, logic, or decoding.

These notes turn progression into a small learning and storytelling system rather than a simple level counter.

Sound and Micro-Animations

I used the browser’s Web Audio API to generate lightweight sound effects without downloading external audio assets.

The game includes:

A rising light chime for correct answers
A low dark tone for wrong answers
A soft Oracle tone for Gemini hints
Golden feedback flashes
Red warning pulses
Blue Oracle glows
Decoder-wheel reactions
Reward sparkles
Button and input micro-interactions

The effects make the machine feel alive and responsive while avoiding copyrighted sound files.

The CSS also respects the user’s prefers-reduced-motion accessibility setting.

Design Direction

The visual identity is built around two opposing energies:

Daylight

Gold
Warm orange
Solar glow
Recovery
Hope

Night and Computation

Deep blue
Black glass panels
Binary patterns
Cold light
Uncertainty

The circular decoder wheel sits between both sides, representing the machine—and the player—trying to keep them in balance.

Challenges I Solved

Making AI part of gameplay

The biggest design challenge was avoiding a normal “click for help” button.

The daylight cost and adaptive modes turned Gemini into a meaningful strategic mechanic.

Protecting the API key

Gemini requests run through a Next.js server route. The key is stored in GEMINI_API_KEY and never sent to the browser.

Keeping transitions safe

Input and action buttons are disabled during level intros and fragment-recovery overlays, preventing duplicate submissions or accidental hint requests.

Flexible answer checking

The answer-normalization system removes formatting differences while keeping the intended solution intact.

Making a small game feel complete

Level intros, archive notes, fragments, sounds, risk tracking, and cinematic endings helped transform a small puzzle prototype into a complete journey.

Prize Category

Best Google AI Usage

Gemini is directly integrated into the game’s central risk-and-reward system.

The Oracle changes its hint style according to the player’s remaining daylight and mistake count. Because every request reduces daylight, AI assistance has an immediate gameplay consequence.

Gemini is therefore not an external helper—it is a dynamic mechanic inside the game.

Best Ode to Alan Turing

The project celebrates Alan Turing through code-breaking, binary systems, algorithms, logic gates, pattern recognition, and machine-inspired reasoning.

The unlockable Turing Notes connect every completed level to the ideas that helped shape modern computing.

What Makes Turing’s Solstice Different

The central idea is simple:

Help has a cost.

The player can ask an advanced AI for guidance, but doing so weakens the resource needed to survive.

Every system supports that idea:

Puzzles test reasoning
Daylight creates pressure
Wrong answers expand the darkness
Gemini offers strategic assistance
Fragments create narrative progression
Turing Notes reward curiosity
Sound and motion make the machine react

The project combines theme, gameplay, artificial intelligence, and presentation into one connected experience.

Final Thoughts

I began with the idea of a single solstice-themed cipher.

It grew into a six-level journey about light, logic, risk, and the process of learning how to decode a machine.

The final light does not return because the Oracle solves everything for the player.

It returns because the player learns when to ask for help—and when to trust their own reasoning.

The machine is waiting.

Can you decode the last light before darkness takes over?

Yousun Agri360: Reviving an Unfinished AgriTech Idea into a 10-in-1 Smart Farming Platform

Yousun Lotif (Bappy Khan) — Wed, 03 Jun 2026 17:45:36 +0000

This is a submission for the GitHub Finish-Up-A-Thon Challenge

What I Built

I built Yousun Agri360, an AI-powered 10-in-1 smart farming platform designed to help small farmers manage their full farming journey from one connected dashboard.

The idea came from a simple problem: small farmers often need to make important decisions about crop health, weather, market prices, finance, selling, credit, and machinery — but these tools are usually scattered, manual, or difficult to access.

Yousun Agri360 brings these workflows together in one place.

Farmers can use the platform to check crop disease support, weather risk alerts, market price insights, farm expense and profit tracking, crop calendar planning, direct selling, credit readiness, Bangla voice support, machinery rental, and smart reports.

The 10 modules are:

AgriDoctor AI — crop disease detection workflow
FarmWeather Guard — crop-specific weather risk warning
AgriMarket Link — market price comparison and best selling option
Farm2Market — farmer-to-buyer marketplace
KrishiBot AI — farming chatbot assistant
AgriCredit AI — digital farmer credit profile
KrishiVoice — Bangla voice assistant concept
FarmLedger AI — farm cost, income, profit/loss, and ROI calculation
CropTime Planner — crop calendar and farming task timeline
MachineryShare Agri — farm machinery rental workflow

To me, Yousun Agri360 is not just a dashboard. It is a foundation for a practical farming operating system that can grow into a real product for small farmers.

Demo

Live Demo: https://yousun-agri360.vercel.app/

GitHub Repository: https://github.com/yousunlotif-bappy/yousun-agri360

Demo Video: https://mega.nz/file/NZ1wnaqa#KbLnth7iDnBa98p6Fh_1-T8Z1Dd7kYnlA8Uof9I8YnQ

Main pages to explore:

Landing Page: https://yousun-agri360.vercel.app/
Dashboard: https://yousun-agri360.vercel.app/dashboard
AgriDoctor AI: https://yousun-agri360.vercel.app/disease-detection
FarmWeather Guard: https://yousun-agri360.vercel.app/weather-alert
FarmLedger AI: https://yousun-agri360.vercel.app/farm-ledger
CropTime Planner: https://yousun-agri360.vercel.app/crop-calendar
AgriCredit AI: https://yousun-agri360.vercel.app/agri-credit
Reports: https://yousun-agri360.vercel.app/reports

The Comeback Story

Before this challenge, Yousun Agri360 was an unfinished agri-tech idea.

I had the problem direction, but the project was not complete. There was no polished dashboard, no connected user journey, no independent module structure, no backend API layer, and no clear before/after story.

The original idea was scattered around real farming problems:

Crop disease detection
Weather uncertainty
Market price confusion
Middleman dependency
Farm cost tracking
Credit difficulty
Machinery access
Low digital literacy

During the Finish-Up-A-Thon, I turned that unfinished idea into a working MVP with a clear product structure.

Before

The project had:

Rough farmer problem notes
No complete dashboard
No module architecture
No working workflows
No backend/API structure
No polished README
No clear submission story

After

Now Yousun Agri360 includes:

A professional landing page
A smart farmer dashboard
10 connected farming modules
Lightweight backend API routes
Functional MVP workflows
A reports page
A public GitHub repository
A live Vercel deployment
A clear before/after completion story

The biggest transformation was turning a rough problem list into a real product experience that users can actually open, navigate, and understand.

Features Added

AgriDoctor AI

Farmers can select a crop, upload a crop image, and receive an AI-style disease diagnosis with confidence score, risk level, possible cause, treatment steps, organic solution, and expert advice.

FarmWeather Guard

Farmers can generate crop-specific weather risk reports with warning level, risk score, recommended actions, avoid actions, and next best action.

FarmLedger AI

Farmers can calculate total cost, total income, profit/loss, ROI, cost per acre, and receive a simple financial recommendation.

CropTime Planner

Farmers can generate a crop calendar with irrigation, fertilizer, disease check, pest check, and harvesting tasks.

AgriMarket Link

Farmers can compare market prices, transport cost, demand level, and net income to identify the best selling option.

Farm2Market

Farmers can add product listings, simulate buyer orders, and track delivery status.

MachineryShare Agri

Farmers can search farm machinery, filter machine type, book machines, start service, and complete service.

AgriCredit AI

Farmers can generate a digital credit profile with credit score, risk level, repayment capacity, loan recommendation, and bank/MFI-style summary.

KrishiBot AI

A text-based farming assistant that detects the farming problem type, gives simple advice, and recommends the correct module.

KrishiVoice

A Bangla voice assistant concept for farmers who may not be comfortable using complex apps or typing long questions.

Backend API Routes

I added a lightweight backend layer using Next.js API Routes.

API Route	Purpose
`/api/agri-doctor`	Crop disease diagnosis
`/api/weather-risk`	Crop-specific weather risk report
`/api/farm-ledger`	Farm finance calculation
`/api/crop-calendar`	Crop calendar generation
`/api/agri-credit`	Farmer credit scoring
`/api/reports`	Platform report summary

The architecture is simple and scalable:

Frontend Pages
↓
Next.js API Routes
↓
Mock AI / Rule-Based Logic
↓
Smart Farming Results
↓
Reports and Dashboard Insights

This makes the project more than a static UI prototype. The main modules send farmer input to backend API routes and display structured farming results.

MVP Limitations

This is a contest MVP, so I kept some parts simulated but honest:

Crop disease detection uses mock AI logic.
Weather risk uses rule-based conditions.
KrishiBot uses rule-based responses.
KrishiVoice is a simulated voice assistant.
Marketplace and machinery booking use local state.
No production database is connected yet.

The architecture is ready for real integrations later.

Tech Stack

Next.js
TypeScript
Tailwind CSS
Lucide React
Next.js API Routes
Mock AI logic
Rule-based decision engine
Vercel deployment

Future Roadmap

Future upgrades include:

Real crop disease ML model
Live weather API
Supabase/PostgreSQL database
Farmer authentication
SMS and WhatsApp alerts
Real speech-to-text and text-to-speech
Payment integration
Bank/MFI dashboard
Marketplace order tracking
Offline-first mobile version

My Experience with GitHub Copilot

GitHub Copilot was a very useful support tool during this project. It helped me work faster and stay more organized, especially when I was writing repeated code and setting up the project structure.

In this project, Copilot helped me create reusable dashboard components, define TypeScript interfaces for module data, and manage form logic and state handling more easily. It also supported me in writing API route boilerplate and preparing mock AI response structures for testing.

Another helpful part was building rule-based decision logic and repeated UI card patterns. These sections required similar coding patterns, so Copilot saved time by suggesting useful code ideas that I could modify according to my project needs.

Copilot also helped me improve the README structure, refactor code, and polish the final project. However, I did not use its suggestions blindly. I reviewed, edited, and customized the code to make sure it matched my project requirements.

The biggest challenge was keeping a 10-in-1 platform easy to understand. To solve this, I designed each module as an independent workflow. A farmer can use only one module, or use the full connected platform.

Final Thoughts

Yousun Agri360 is not just another agriculture app. It is a modular farmer operating system where each service works independently, but together they support the complete farmer journey from planning to production, selling, finance, and machinery access.

DEV Community: Yousun Lotif (Bappy Khan)

Turing’s Solstice by Yousun — An AI-Powered Puzzle Game Where Every Hint Costs Light

Turing’s Solstice by Yousun

What I Built

How It Connects to the June Solstice

Video Demo

Live Demo

Code

How I Built It

Core Game State

Google Gemini as the Solstice Oracle

Cryptic Mode

Balanced Mode

Urgent Mode

Reliable Fallback System

Fragment Recovery System

Alan Turing Inspiration

Sound and Micro-Animations

Design Direction

Daylight

Night and Computation

Challenges I Solved

Making AI part of gameplay

Protecting the API key

Keeping transitions safe

Flexible answer checking

Making a small game feel complete

Prize Category

Best Google AI Usage

Best Ode to Alan Turing

What Makes Turing’s Solstice Different

Final Thoughts

Play the Game

Watch the Demo

Explore the Code

Yousun Agri360: Reviving an Unfinished AgriTech Idea into a 10-in-1 Smart Farming Platform

What I Built

Demo

The Comeback Story

Before

After

Features Added

AgriDoctor AI

FarmWeather Guard

FarmLedger AI

CropTime Planner

AgriMarket Link

Farm2Market

MachineryShare Agri

AgriCredit AI

KrishiBot AI

KrishiVoice

Backend API Routes

MVP Limitations

Tech Stack

Future Roadmap

My Experience with GitHub Copilot

Final Thoughts