SleepFix

Inspiration

The inspiration for this app came from the constant frustration and wasted time developers face when debugging runtime errors in web applications. Our goal was to create a tool that not only catches errors but also fixes them autonomously, reducing the manual effort involved in bug fixing and allowing developers to focus on building features.

What it does

The app monitors your web application in real-time for runtime errors. When an error is detected, it uses LLM agents to analyze the error and inspect the codebase to pinpoint the cause of the issue. Once identified, the app generates a fix and creates a pull request with the proposed changes, saving developers significant time in the debugging and fixing process. It acts as an autonomous team member that continuously ensures the stability and reliability of your web application.

How we built it

We are using a combination of multiple agents using Prem and Groq to work together to find the cause and create the github pull request with the fix and the description of the issue.

We are using Fetch AI as an agent to observe and monitor github pull requests and create Slack or email notifications to the user when there is a bug fix available.

We are hosted on Netlify and using Vectara AI for Rag to search the code.

Once identified, the app generates a fix and creates a pull request with the proposed changes, saving developers significant time in the debugging and fixing process. It acts as an autonomous team member that continuously ensures the stability and reliability of your web application.

Challenges we ran into

One of the biggest challenges was ensuring the accuracy and relevance of the fixes suggested by the LLM agents. We had to tune the prompts and the models extensively to understand the context of the code and to make appropriate fixes. Mapping a runtime bug to a real code file in a github repository was also a challenge requiring parsing source map files on the client side.

Accomplishments that we're proud of

We have the full prototype working end to end and manages to pull off everything we set out to do for the hackathon.

What we learned

Throughout the development process, we learned a great deal about the limitations and potential of LLMs in software development. We learned about how to combine all of these technologies and combine them to create something genuinely great and useful.

What's next for SleepFix

Next, we plan to expand the capabilities of the app to automatically test the fix that it created to ensure the bug genuinely is fixed. In which case it could also automatically deploy the fix into production.

Built With

• fetchai
• groq
• netlify
• prem
• python
• typescript
• vectara