Inspiration
I recently got scammed when I purchased something from a website that appeared legitimate. A deeper search revealed several red flags. It's easy to forget to research before buying or investing. That's why I created this extension, to analyse websites while you browse, helping save time, money, and protect against scams. Using Perplexity's Search API and the Sonar-Pro model, the extension assesses reviews, scam reports, sentiment, complaint patterns, and fake review indicators to deliver an overall risk score.
What it does
ScamDetector is a browser extension that automatically analyzes any website you're browsing. It searches the web for reviews, scam reports, sentiment patterns, and indicators of fake content to provide a risk score in real-time, helping you make safer decisions online.
How we built it
I built the extension using JavaScript and integrated it with a Flask backend. The backend calls the Sonar-Pro model and interprets the results to generate a structured risk assessment. We also used CORS for secure communication between frontend and backend.
Challenges we ran into
- Parsing unstructured data and transforming it into a meaningful risk score.
- Ensuring consistent JSON responses from the LLM.
- Handling CORS policies for smooth frontend-backend interaction.
- Designing an intuitive yet non-intrusive UI for the extension.
Accomplishments that we're proud of
- Successfully integrated Perplexity's search capabilities with a custom risk analysis prompt.
- Built a working browser extension that operates in real-time.
- Designed structured output using Pydantic models.
- Created a full-stack system with automatic website analysis.
What we learned
- How to structure LLM prompts to yield consistent JSON outputs.
- Leveraging search-based APIs for real-time data collection and analysis.
- Developing browser extensions and managing cross-origin communications
What's next for ScamDetector Extension
- Integrate a chatbot to answer user questions in real-time. The users dont have to go somewhere else for further search
- Enhance UI/UX for better accessibility and experience.
- Include an option to display in-depth reviews and tailored next steps, especially if the user suspects they’ve been scammed.
- Consider integrating a chatbot to offer real-time assistance, so users can ask follow-up questions without needing to leave the page.
Log in or sign up for Devpost to join the conversation.