TrolleyRush

• 

  TrolleyRush Thumbnail

• 

  Main Splash Screen featuring clean and responsive interface

• 

  Personalized profile verification node celebrating solo-architect deployment checkpoints

• 

  Live implementation dashboard displaying choosing customized username/avatar

• 

  Dashboard showcasing levels and challenges(level 1-12 are free and rest upto 35 are paid)

• 

  Real time Gameplay and score calculation metrics

Inspiration-

TrolleyRush was inspired by classic high-velocity endless runners and survival arcade games like Subway Surfers and Temple Run. The goal was to take those core continuous-movement mechanics and re-engineer them into a focused, level-based delivery objective with a unique, comedic narrative wrapper.

What it does-

The game follows a quirky baker navigating a chaotic track with a customized trolley/cart. Players must maintain momentum and dodge obstacles to safely reach the destination storefront, "CAKE BAKE," ensuring time-sensitive pastry deliveries to awaiting customers. The core gameplay loops combine physics-based movements, obstacle avoidance, and precise timing.

How we built it-

As a solo developer, I utilized a rapid AI-assisted development workflow (vibe coding) to spin up the architecture.

• Core Generation: Leveraged freebuff.ai using the structured CRAFT prompting model to establish the initial codebase framework.
• Refinement & Iteration: Used bolt.new for rapid codebase modifications, hot-reloading component patches, and UI/UX fine-tuning.
• Deployment: Hosted the live distribution build on Vercel's edge network for zero-configuration global hosting.

Challenges we ran into-

The primary bottleneck was managing LLM hallucinations and structural code regressions during multi-feature prompt injections. The AI would frequently introduce syntax errors or drop existing gameplay event listeners when adding new components. I overcame this by breaking down features into isolated modular blocks and applying highly specific structural parameters to keep the context window stable.

Accomplishments that we're proud of-

I am incredibly proud of launching a fully functional, end-to-end interactive game loop as a solo developer with zero manual syntax roadblocks. Furthermore, I successfully distributed the live prototype to a public test group of peers and friends, receiving excellent user feedback and engagement metrics right out of the gate.

What we learned-

This project served as an advanced deep-dive into high-fidelity Prompt Engineering. I mastered the CRAFT model framework, learning how to engineer deterministic system prompts that guarantee scalable code output from LLMs under tight deadlines. Additionally, I gained practical experience in product monetization and ad-ops by successfully integrating the Monetag SDK wrapper to inject programmatic ad revenue modules directly into the live web build.

What's next for TrolleyRush-

With the core architecture of TrolleyRush stabilized, I am transitioning my rapid prototyping pipeline toward a health and wellness ecosystem. The conceptual framework and high-fidelity UI layout for a physical and mental status tracking application are already locked in; the next sprint will focus on integrating predictive LLM models to analyze user logs.

Built With

• adops
• bolt.new
• freebuff
• generativeai
• html5
• javascript
• monetag
• promptengineering
• vercel
• vibecoding