auto-sre

Inspiration

The inspiration came from the critical need to automate bug detection and resolution in production systems. Traditional SRE (Site Reliability Engineering) relies heavily on manual debugging, which is time-consuming and error-prone. We wanted to create an AI-powered system that could automatically reproduce bugs, analyze code, and suggest fixes - essentially automating the entire bug hunting and fixing process.

What it does

Auto-SRE is an intelligent bug hunting and resolution system that combines three powerful technologies: Browser Automation: Uses Browser Use to automatically reproduce UI bugs by following step-by-step instructions Code Testing: Leverages Daytona sandboxes to test potentially buggy code in isolated environments AI Analysis: Employs Google Gemini to analyze investigation results and suggest precise fixes The system can take a bug report, automatically reproduce the issue, test the suspected code, analyze the root cause, and provide actionable fixes with test cases.

How we built it

Frontend: React-based web interface for submitting bug tickets with a modern, intuitive UI Backend: Python Flask API with real-time WebSocket communication for live investigation updates Browser Automation: Browser Use library for headless browser automation and screenshot capture Code Testing: Daytona API integration for secure code execution in isolated sandboxes AI Integration: Google Gemini API for intelligent bug analysis and fix suggestions Infrastructure: Docker Compose for microservices architecture with separate web, API, and gateway services Tech Stack: Python, React, Flask, Socket.IO, Browser Use, Daytona, Google Gemini, Docker

Challenges we ran into

Browser Automation on macOS: Initial issues with browser permissions and headless mode configuration API Integration Complexity: Coordinating multiple APIs (Browser Use, Daytona, Gemini) with different authentication methods Real-time Updates: Implementing WebSocket communication for live investigation progress Code Analysis Accuracy: Ensuring AI-generated fixes are accurate and testable Environment Setup: Managing multiple services and ensuring proper communication between components

Accomplishments that we're proud of

Successful Bug Reproduction: The system successfully reproduced a critical payment bug (coupon code issue) that was charging full price instead of discounted amount End-to-End Automation: Complete automation from bug report to fix suggestion Real-time Monitoring: Live updates during investigation process with detailed logging Accurate Root Cause Analysis: AI correctly identified the backend code issue and provided precise fix Cross-Platform Compatibility: Resolved macOS-specific browser automation challenges Scalable Architecture: Microservices design allows for easy scaling and maintenance

What we learned

Browser Automation Complexity: macOS security permissions require careful configuration for headless browser automation AI Integration Best Practices: Proper prompt engineering is crucial for accurate code analysis Real-time Communication: WebSocket implementation requires careful error handling and connection management Multi-API Coordination: Successfully orchestrating multiple external APIs requires robust error handling User Experience: Real-time feedback during long-running investigations significantly improves user experience

What's next for auto-sre

Enhanced AI Models: Integration with more advanced AI models for better code analysis Automated Fix Deployment: Direct integration with CI/CD pipelines to automatically apply fixes Multi-Language Support: Extend beyond JavaScript/Python to support more programming languages Performance Monitoring: Add real-time performance monitoring and alerting Machine Learning: Implement ML models to learn from past bug patterns and improve detection accuracy Enterprise Features: Role-based access control, audit logging, and compliance reporting Mobile App Testing: Extend browser automation to mobile app testing Integration Ecosystem: Connect with popular bug tracking systems (Jira, GitHub Issues, etc.)

Built With

• browseruse
• cursor
• daytona
• gemini
• javascript
• python
• typescript