SPOOKYTIME

Inspiration

I wanted to turn a simple services marketplace into something playful and eerie—an experience where people could buy and sell their time inside a neon-haunted Halloween interface. The idea of adding AI-generated trust scores made the interactions feel alive and unpredictable.

What I Learned

I learned how Clerk JWTs work with Supabase RLS, how to fix UUID/text mismatches, how to configure secure storage uploads, and how Kiro’s spec-driven workflow speeds up building complex UIs and APIs.

How I Built It

SpookyTime uses Next.js, Clerk, NeonDB Postgres, Supabase Storage, and an AI trust-score engine. Kiro generated the skeleton, components, and theme, while I implemented API routes, RLS rules, database schema, and AI-powered scoring.

Challenges

I faced issues with database connection errors, RLS policy failures, Clerk JWT configuration, and storage uploads. Fixing these taught me how modern auth, serverless compute, and Postgres security all fit together.

Built With

• next.js
• supabase