It's purely client-side aggregation. The entire "engine" is a single HTML file with vanilla JavaScript—no backend server required. When you type a query, it fires off parallel fetch() requests (using Promise.all) to public, CORS-friendly APIs like Algolia (for HN), Reddit’s public .json endpoints, and Wikipedia’s API. It then normalizes those JSON responses and injects them into the DOM. History and settings are just persisted in localStorage.