GeoLocator — AI-Powered Photo Geolocation

Upload a photo, pay 100 sats via Lightning Network, and Gemini 2.5 Flash identifies the top 3 most likely locations the photo was taken.

Stack

• Next.js 16 (App Router)
• @moneydevkit/nextjs — Lightning Network checkout loop
• Google Gemini 2.5 Flash — geospatial-intelligence analysis
• shadcn/ui — component design
• Vercel — deployment

Setup

1. Clone and install

git clone <repo>
cd mdk
npm install

2. Environment variables

cp .env.local.example .env.local

Fill in:

Variable	Where to get it
MDK_ACCESS_TOKEN	moneydevkit.com/dashboard or npx @moneydevkit/create
MDK_MNEMONIC	Same as above
GEMINI_API_KEY	aistudio.google.com/app/apikey

3. Run locally

npm run dev

For local Lightning payments you need to expose your dev server:

ngrok http 3000

Then set your app URL in the MDK dashboard to https://<your-ngrok-id>.ngrok-free.app.

Deploy to Vercel

vercel deploy

Add the three environment variables (MDK_ACCESS_TOKEN, MDK_MNEMONIC, GEMINI_API_KEY) in your Vercel project settings, then set your app URL in the MDK dashboard to your Vercel deployment URL.

Privacy

Images are never stored server-side. The uploaded photo is held in sessionStorage for the duration of the checkout redirect and deleted from the browser immediately after the Gemini analysis completes.