Inspiration
Picture this: a majestic 6'3" Chinese boy, dripped out in jorts and a button-up shirt, carrying a Harvard tote bag, wired headphones in, walking around campus while reading Meditations by Marcus Aurelius.
Oh shoot — sorry, I was describing myself.
Anyways... inspired by "starter pack" memes that fuel internet culture and flood our Instagram feeds, we built Performeter to answer the question: how performative is this person, really? The result is a playful — but surprisingly insightful — tool that makes the Tinder experience more self-aware (and fun).
What it does
Performeter is a Chrome Extension that:
- Automatically parses data from Tinder profile bios and (optionally) detects objects and fashion in profile images
- Uses Gemini to understand what a "performative male" is and what traits constitute being "performative"
- Produces four category scores:
- Trendy — references to cultural objects like matcha, tote bags, film cameras
- Personality — curated tagline-style bios
- Aesthetic — level of profile curation and photo staging
- Buzzwords — meme energy, TikTok slang, and internet culture density
- Displays an overall score and a clean, gamified UI overlay with a category breakdown
How we built it
- Chrome Extension (Manifest V3) for injecting scripts and rendering a sidebar overlay
- Content Script scrapes bio text and profile info
- Background Script calls the Gemini/Claude API with a structured prompt and rubric
- Frontend UI built with HTML/CSS and injected dynamically into the Tinder DOM
- Design prototyped in Figma, refined collaboratively, then implemented in Cursor
- Testing with dummy JSON responses before connecting to live AI models
Challenges we ran into
- DOM scraping on Tinder/Hinge was difficult due to dynamically loaded, obfuscated elements
- Getting AI to produce valid JSON consistently — solved with careful prompt engineering
- Balancing speed (API latency) with UI smoothness to avoid a janky user experience
- Styling an overlay that works across different screen sizes and dark/light modes
Accomplishments that we're proud of
- Fully working Chrome extension built in less than 5 hours (shortest hackathon ever — wth!)
- Clean, responsive UI that matches the meme energy of the project (yes, we used Comic Sans — unironically)
- Successfully integrated AI that returns consistent structured output
- Created a playful experience that is fun to demo live and works reliably
What we learned
- How to structure Chrome extensions using Manifest V3
- Prompt engineering for consistent, structured AI outputs
- Rapid design-to-code iteration using Figma MCP + Cursor
What's next for Performeter
- Add computer vision to detect a wider range of visual performative cues (matcha cups, film cameras, etc.)
- Expand support to Bumble, OkCupid, and other dating platforms (?)
- Allow user customization (choose which categories matter most — maybe even detect performative sub-genres?)
Built With
- css
- html
- javascript
- manifestv3
- react
- vite

Log in or sign up for Devpost to join the conversation.