RepoScan

Inspiration

Last month, a doctor missed a critical finding on my mother's X-ray. By chance, a specialist in our family caught it and she was rushed to emergency oxygen.

What it does

RepoScan analyzes X-rays, MRIs, and skin photos using AI and returns a structured clinical report in plain English — with urgency detection, immediate action steps, and direct contact to the right specialist from a curated directory of 22 top doctors in Kashmir Valley.

How we built it

React + Tailwind CSS frontend, powered by OpenRouter Vision AI for multimodal scan analysis. Engineered a professional master prompt for structured clinical output. No backend, no database — fully client-side. Built using Claude Code in one hackathon day.

Challenges we ran into

Gemini 1.5 Pro was deprecated mid-build, returning 404 errors. Pivoted to OpenRouter's free vision API, which required restructuring the entire request format. Prompt engineering for honest yet calm clinical output also took several iterations.

Accomplishments that we're proud of

A production-ready medical tool built in one day — real doctor directory, real urgency detection, real clinical reports. It actually works on a real X-ray from a real patient.

What we learned

How to engineer multimodal vision prompts for structured medical output, handle API deprecations mid-hackathon, and build a complete React app with routing and animations using Claude Code.

Built With

• react
• tailwindcss