Hintify

Inspiration

Debugging can be frustrating, especially when you’re stuck on a logical bug. We wanted an AI assistant that doesn’t just give answers but nudges you in the right direction—like a mentor would.

What it does

Hintify reviews your code and identifies major logical bugs, giving you hints instead of direct solutions. We make it even more fun by having characters review your code in their unique style! So you now have a buddy to debug your code with.

How we built it

We built Hintify using JavaScript and the VS Code extension development support. Hintify utilizes GPT-3.5 APIs to analyze active files and search for logical errors. To enhance interactivity, we employed sox to record the user's voice input and transcribe it using the whisperAPI. The whisper calls enable us to process even multilingual user data. After transcription, the query runs through GPT-3.5 and is fed to the eleven_flash_v2_5TTS model by ElevenLabs.

Challenges we ran into

We wanted to make the extension robust and interactive so people can leverage the potential of pair programming. So, we gave special attention to adopting different characters' personas so that the debugging process could be more engaging. Additionally, we wanted to make sure that our output is not giving the users direct solutions or code but instead giving them hints around their queries.

Accomplishments that we're proud of

We were successful in finishing the product we initially thought of. As a cherry on top, we were able to implement a Mickey-mouse voice in our code assistant so people can have it while coding.

What we learned

The whole team learned a lot about managing multiple API calls (a lot of callbacks). Additionally, we got to work with voice APIs like whisper and Elevenlabs and explored their potential. (we didn't know whisper can support 98 languages!)

This was also our first time making a VS code extension so it was great to know how things get built inside our fav code editor.

What's next for Hintify

We want to push this extension on the marketplace, but seeing how many local dependencies it uses, we will need to figure out a way around that. We plan to add more characters and the ability to switch between them, support additional languages, and introduce gamification elements to make debugging even more engaging.