Inspiration

While it started off as a joke suggestion for a "predator catching app" where decoys talked with predators to gather their info to send off to the police as evidence, we quickly realized that such an idea was not realistic as once word got out that our app was a trap, then predators would no longer use it. So, we brainstormed other ideas, keeping the fundamental idea of taking screenshots and preventing sketchy behavior online, and took inspiration from other Chrome extensions like GoGuardian and Bark to come up with SafetyGuard.

What it does

SafetyGuard monitors the content on any website you visit, and alerts the user when dangerous content has been found. Instead of blocking the content completely, it serves as an informative warning to a younger user, since it's easier to know to avoid something if you understand why it's problematic.

How we built it

The project was built from the ground up using Google Gemini to transform our ideas and concepts into workable code. Additionally, art and images were manually drawn and edited using Turbowarp, and integrated into the program with more Gemini code.

Challenges we ran into

Some challenges we ran into included only one person being able to edit a file at a time on GitHub to avoid overwrite conflicts, vital team members leaving for extended periods of time, and the results of other non-Gemini Chat Bots making errors and being unhelpful at programming.

Accomplishments that we're proud of

Functionality. We're most proud that the extension actually works, you can download it, pin it to your top bar, and it'll scan for red-flag keywords on any site you visit and alert you of them, just as we planned. We're also proud of the artwork, the little drawings to go along with the actual functioning extension makes it feel professional, like something you'd find on the Chrome Web Store.

What we learned

Google Gemini is really good at coding! Github Copilot is not.

What's next for SafetyGuard

As of right now, SafetyGuard is a pretty basic text scanner; it doesn't take context into account, and it doesn't know how to parse subtleties, or innuendos. In the future, we would love to get an AI to both analyze the text for hidden meaning, and to generate a unique Safety Tip, but that'll be saved for another time.

Built With

Share this project:

Updates