Inspiration

Want to find a way to check code that's vulnerable using LLM while being effective, so we created the multi-agent system

What it does

Uses many nodes of agents to discuss whether pieces of code are vulnerable

How we built it

Using langchain, next.js, and fastAPI

Challenges we ran into

Trying to convert Latex into PDF and detecting languages

Accomplishments that we're proud of

Was able to get websockets to work from backend to frontend, real-time communication

What we learned

Learned how multi-agent systems works and its benefits over single-agent systems

What's next for CodeSentinel

Get PDF to work and be able to handle larger pieces of code and even repositories by learning chunks and data transformation

Built With

Share this project:

Updates