-
-
LogicHinter
-
QuizCraft
Inspiration
I wanted a tool that helps people think, not one that simply hands out answers or teaches how to "copy". Most AI chats and tutorials solve problems directly, but I needed a brainstorming buddy that teaches how to reason. That idea became LogicHinter. Then I expanded the same concept for students and created StudyHinter—a playful, adaptive study companion.
What it does
LogicHinter guides users through code algorithm and gives insightful links once a correct option is chosen - without giving away solutions. StudyHinter generates quizzes, images, explanations, and difficulty-adaptive learning paths. Both apps turn AI into a structured, hint-driven teacher rather than a solution generator.
How I built it
I built a reusable FastAPI + React skeleton, added Gemini reasoning engines, designed strict prompt/validation pipelines, and wrapped everything in spooky, atmospheric Halloween UI. Two full applications—LogicHinter and StudyHinter—were generated from one clean architectural skeleton.
Challenges I ran into
Ensuring AI never reveals full solutions
Keeping JSON outputs structured and reliable
Integrating image enrichment and fallbacks
Designing two distinct apps from one codebase
Matching the Halloween theme without harming usability
Accomplishments that I'm proud of
Built two complete apps from one skeleton
Created a hint-only reasoning engine that actually teaches
Designed polished spooky UI with smooth interactions
Achieved consistent, production-ready AI behavior
Delivered a versatile system usable for coding, studying, and more
WhatI learned
Specs + strict prompts create predictable AI outputs
Reliable AI apps need validation layers and fallbacks
One skeleton can support multiple, fully unique applications
Good design enhances clarity—even with spooky themes.
Kiro works well when provided with screenshots and textual prompts.
What's next
Add more subjects and learning modes.
Turn LogicHinter into a full coding-interview coach.
Add multiplayer quizzes for StudyHinter.
Expand the skeleton into a general “hint-based learning engine”.
Built With
- github
- javascript
- kiro
- netlify
- python
- railway
- react
- tailwind-css
- typescript
- vite
