Inspiration
As a full-stack programmer myself, I've run into many situations where after doing a pull, there are sudden new issues that came up due to new commits. Especially in full-stack projects, a small change in a specific file can have a domino effect which causes the entire app to run into issues. Thus, I believed such a tool would have helped me and my teammates greatly in avoiding unnecessary time lost on debugging/identifying root causes.
What it does
This tool automates the process, utilising a Github link and provided timeframe to highlight potentially problematic commits, and suggesting rollbacks to help developers constrict their scope to look through specific commits and the associated changes, rather than having to look through multiple.
How I built it
I used Kiro to first build a full-stack app based on the idea, then tweaked the frontend and backend features according to what suited the idea best.
Challenges I ran into
I first tried to implement the AI reasoning with AWS Bedrock, but ran into some issues with enabling the token quota. Thus, after trying with it for a while, I shifted to use GroqAPI instead, and use Meta's Llama reasoning model.
Accomplishments that I'm proud of
I'm proud that I was able to apply my previous internship knowledge of cloud deployment to quickly deploy my application, and quickly identify and resolve issues I was familiar with. This showed me how far I've come since my first attempt at deployment, which took me much longer to properly figure out.
What I learned
I learned more about various AWS services, such as Bedrock and Elastic IPs.
What's next for incident-tracker
More secure methods of data storage and retrieval, such as Postgres. Also using a local RAG LLM instead of making requests to a model online, in case private repository information is sensitive.
Suggested Use
To test it with a public Github repository, try it with https://github.com/langflow-ai/langflow and a date range of 29/6/2026 00:00 to 3/7/2026 00:00. You can also test it with one of your private repositories as long as you provide your Personal Github Token (creation available at github.com/settings/tokens)
Log in or sign up for Devpost to join the conversation.