Phish Buster

manually designed and colored logo

Inspiration

Spam is getting smarter. AI-generated phishing emails look more convincing than ever, and the people that are most at risk are older adults and non-technical users who can't spot the fabrication. We wanted to create something simple enough that anyone could use without a tutorial.

What it does

Upload a screenshot of a suspicious message or paste a URL. PhishBuster runs it through an AI analysis layer, then returns a color-coded verdict with a confidence breakdown explaining why, specific red flags, easy to understand advice on what to do next, and a one-tap share button so you can warn someone you care about.

How we built it

HTML, CSS, and JavaScript. The AI analysis runs through the Groq API. We used Node.js as a local development server for testing, and deployed to GitHub Pages for production. Claude assisted with streamlining the development process. The logo was a manual process: Pinterest for inspiration, Canva and ChatGPT for generation, then hand-colored in Pigment.

Challenges we ran into

API integration was a steep learning curve, we hit rate limits, burned through free tier quotas debugging, and had to swap API providers mid-build before finding a stable configuration. AI image generation kept producing generic output for the logo, so we went manual and made something we actually liked.

Accomplishments that we're proud of

First hackathon!! One brought technical skills, the other brought design taste and presentation, combining these two to make a useful, real world product that tackles a real problem. First time integrating an AI API into a real project. Got over the fear of demoing in front of people. Built a clean, custom UI with a personalized logo and custom domain.

What we learned

The full lifecycle of shipping a web product: design, frontend, API integration, deployment, debugging production issues under time pressure, and presenting it. How to scope for a 24 hour build and knowing what to cut when time is tight.