Inspiration

Many CS projects nowadays are built off the back of AI agents like Claude or ChatGPT without the programmers actually understanding how their agent changed the program or how to make changes on their own. We were inspired to fix this problem and develop a Gemini powered app which would help improve users understanding of the code implemented by their agents.

What it does

The app we created reads a code base and an AI agents chat history and uses it to create tools to help users learn about the new code. The app can generate flash cards to help you study how the new code works as well as quizzes that you can use to test your knowledge. It will then show you how well you scored on the quiz later and can generate new quizzes whenever you add new code. We also created this app as a VS Code extension, so users preferring VS Code can install this extension and run it while programming, while fans of other IDEs can run the app on another tab.

How we built it

We used AI to help with the design process, but used node.js and react to complete the app, using Gemini as our agent which reads chat history and generates the quizzes. We used the extension Builder to help write the VS Code extension.

Challenges we ran into

As usual in a hackathon, we had to crunch at the end in order to finish up both the VS code extensions and the App at the same time. We were able to overcome this by splitting up our work and letting our most talented programmers show of their skills.

Accomplishments that we're proud of

We managed to complete two projects simultaneously in an extremely short time frame, only about 6 hours. This is an extremely difficult task for most talented groups, and we're proud that we were able to accomplish so much in such little time.

What we learned

We learned how to better communicate as a team, how to delegate tasks, and how to efficiently use Gemini to solve problems in the real world. By understanding the power of Gemini's API, we learned how to best utilize to help others without simply creating a wrapper for an AI agent.

What's next for VEX

VEX can still be improved and iterated on in many ways, such as improving the quality and depth of our questions. For example, VEX could be updated to ask questions to help you prepare for interviews, or used as a tool to help beginner programmers still learning how to code the basics. It could also come up with tutorials to help you understand libraries you've never worked with or help you understand teammates posts on GitHub.

Built With

Share this project:

Updates

Submission history