frontpage of xwrapped
project flow

X Wrapped - Devpost Submission

Inspiration

We wanted to bring the viral "Spotify Wrapped" experience to Twitter/X. Everyone loves seeing their year in review as a shareable, cinematic piece of content—but there's no equivalent for your Twitter presence. We built X Wrapped to let anyone generate an AI-powered personality analysis and cinematic video from their tweet history.

What it does

X Wrapped takes any Twitter/X handle and generates:

AI Personality Analysis — Uses OpenRouter to analyze tweet content and determine personality archetypes, top topics, sentiment, and mood
Web Context Enrichment — Optionally uses Exa neural search to pull public context about the person
Cinematic Video — Uses MagicHour's API to generate a shareable 5-second video based on the personality analysis
Voiceover — Generates AI voiceover audio describing the personality
Results Page — Displays the video, stats, and download/share options

All data is stored persistently in MongoDB Atlas.

How we built it

SvelteKit + Svelte 5 — Modern reactive web framework with runes
TypeScript — Full type safety
TailwindCSS v4 — Dark-themed, mobile-first UI
MongoDB Atlas — Persistent storage for wrapped jobs, profiles, and analysis results
OpenRouter — Free AI model for personality archetype analysis
MagicHour — Text-to-video and image-to-video generation
Exa — Optional neural web search for public context
twitterapi.io — Optional live tweet fetching

The pipeline orchestrates: scraping → analysis → video generation → storage → results display.

Challenges we ran into

SvelteKit deployment was the biggest challenge. Getting the app production-ready required careful configuration for:

Environment variable handling between server and client
MongoDB connection management in a serverless-style deployment
Proper handling of async job processing
Video generation polling and webhook handling

We also had to work around MagicHour's prompt length limits (2000 chars), handle profile image uploads from Twitter's CDN, and gracefully fall back when APIs were unavailable.

Accomplishments that we're proud of

Successfully integrating multiple APIs (OpenRouter, MagicHour, Exa, MongoDB) into a cohesive pipeline
Building a responsive dark-mode UI that feels native to X
Creating a robust fallback system so the app works even with minimal API keys
Implementing video backfill so jobs can regenerate media after the fact

What we learned

SvelteKit's deployment model requires careful planning for async server-side operations
MongoDB is excellent for storing flexible job states with varying fields
MagicHour's video generation works great for mood/palette-driven content but has prompt constraints
Building for multiple API providers requires solid error handling and fallback logic