Korean Food Arena

• 

  main

• 

  leaderboard

Inspiration

Food opinions are universal—and endlessly debatable. I wondered: could I ship a playful, data‑driven way to compare Korean dishes in a single day using Kiro’s Spec‑Driven Development?

What it does

Korean Food Arena shows two dishes; you pick one, drop a short comment, and optionally tag your nationality. The leaderboard updates with ELO so everyone sees which dishes are trending and how often they’re chosen.

How we built it

I used Kiro’s spec→design→tasks pipeline. After I described the goal, Kiro drafted the requirements, design, and task list. Hitting “Start Task” iteratively produced code and tests. The app runs as a React + TypeScript frontend with a Cloudflare Worker backend and Drizzle‑managed SQL. The ELO update is standard: $$ E = \frac{1}{1+10^{\Delta/400}},\quad R' = R + K(S - E). $$

Challenges we ran into

• Small edits sometimes felt less fluid in Vibe Coding, so I leaned on the spec flow.
• Balancing ELO responsiveness vs. stability; protecting against vote spam.

Accomplishments that we're proud of

• Spec‑to‑prototype in a day, with automated tests along the way.
• Smooth mobile UX and a leaderboard that “feels fair.”
• Simple data model that supports comments and nationality insights.

What we learned

• Strong, testable specs are the best fuel for AI coding agents.
• ELO fits pairwise food rankings and communicates results intuitively.
• Tiny UX details (copy, spacing, feedback) meaningfully increase engagement.

What's next for Korean Food Arena

• Rich insights: per‑country views, trend lines, and head‑to‑head breakdowns.
• Better moderation and optional accounts.
• Localization, accessibility, and social share cards.

Built With

• cloudflare
• d1
• kiro
• react
• workers