BugReplay: Helping Developers Learn from Bugs Their Team Has Already Solved 🐛

This submission is for the Hacktoberfest Weekend Challenge, titled "Build for a Friend." Debugging can be an arduous task, and sometimes it seems like your teammate has already solved the exact same bug you're facing. That's where BugReplay comes in. I'm developing an AI-powered debugging assistant to transform past bug fixes into reusable knowledge for teams.

Instead of sifting through old chat logs, GitHub issues, or asking teammates for the same solution, developers can simply input an error message or stack trace into BugReplay. It will then locate relevant bugs that the team has already resolved. The system utilizes semantic search to find similar issues, provides insights into root causes, offers AI-powered troubleshooting suggestions based on verified solutions, and saves newly resolved bugs so the entire team can benefit.

It also fosters a collaborative environment where teammates can independently debug, while mentors can review and verify solutions.

BugReplay is built using React for the frontend, Python and FastAPI for the backend, Qwen, an open-weight model for AI, PostgreSQL with pgvector for the database, and employs Retrieval-Augmented Generation (RAG) for the AI workflow. The motivation behind this project is to make debugging a shared learning experience, ensuring that the entire team can benefit from the knowledge of previously fixed problems.

The goal is to reduce repeated debugging, offer more context, and help developers solve problems with greater understanding. I'm thrilled to work on this project with my teammates, exploring how open-weight AI can address real-world issues for development teams. You can find the code in the GitHub repository: https://github.com/jhashivam0022/bugreplay.git. More updates and a demo will be available soon!