# Gemini World

- **Event:** [Gemini 3 ハッカソン 東京](https://cerebralvalley.ai/e/gemini-3-tokyo-hackathon)
- **When:** Sat, Feb 21 at 9:00 AM – 10:00 PM (GMT+9)
- **Where:** 東京都渋谷区渋谷 3 丁目 21 - 3 渋谷ストリーム 5F
- **Team:** [Taiki Nakamura](https://cerebralvalley.ai/u/taylor259), [Yuki Murakami](https://cerebralvalley.ai/u/muchan23), [Malachi Hul](https://cerebralvalley.ai/u/mrmalachi)
- **GitHub:** https://github.com/muchan23/gemini_world
- **Demo video:** https://drive.google.com/drive/folders/16CNwUF2EOla0qP3e2hSdfd0HHQiQm7i9?usp=drive_link
- **Gallery:** https://cerebralvalley.ai/e/gemini-3-tokyo-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/gemini-3-tokyo-hackathon/hackathon/gallery/26

GeminiWorld is a web-based, explorable 2D world where AI-driven minigames reshape the environment in real time via Google’s Gemini API. Players move through a tile-based overworld, enter zones to play minigames, and see terrain, NPCs, weather, and story change based on performance.

Core loop: play minigame → Gemini evaluates → world mutates.

Minigames are triggered by interactions (e.g., hitting a box) and are personalized per player. Gemini generates unique challenges, adapts difficulty, and keeps each run distinct.

Stack: Phaser + React + TypeScript + Vite, deployed on Vercel/Cloudflare Pages.
Team (3): World/Shell, Minigame Framework, AI/Gemini Integration.
Differentiator: Gemini is the backbone—adaptive difficulty, procedural NPC dialogue, emergent storylines, and Live API (voice/camera) enable deep AI-native gameplay.

---

Markdown version of https://cerebralvalley.ai/e/gemini-3-tokyo-hackathon/hackathon/gallery/26. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt