🚀 Inspiration
Modern cloud deployments fail silently or require hours of manual debugging when crash logs pile up. Developers spend countless hours parsing through stack traces to fix minor bugs. We built DevPulse AI to create an autonomous, self-healing pipeline that bridges crash logs directly to AI-powered code patches and automated GitHub Pull Requests.
🛠️ What it does
DevPulse AI acts as an autonomous code-to-cloud diagnostic agent:
- Log & Trace Parsing: Automatically scans terminal output and stack traces.
- AI-Powered Root Cause Analysis: Uses LLM intelligence to pinpoint the exact broken file and line number.
- Self-Healing Code Generation: Generates precise, production-ready patch code with robust exception handling.
- Automated GitHub Deployment: Creates pull requests directly on the repository with zero human friction.
⚙️ How we built it
- Python CLI Architecture: Built using
typerandrichfor a clean, developer-friendly terminal interface. - AWS Bedrock / LLM Diagnostics: Leveraged AI models to analyze complex stack traces and suggest safe patches.
- GitHub API Integration: Automated branch creation and PR submission directly to the code repository.
🚧 Challenges we faced
Structuring terminal log parsing so the AI receives clear context without noise was a challenge. We optimized log truncation and contextual prompt engineering to ensure accurate diagnosis every time.
💡 What we learned
Building developer tools with self-healing capabilities can save thousands of hours in CI/CD pipelines. We learned how to seamlessly connect CLI logging with automated Git workflows.
🔮 What's next for DevPulse AI
- Real-time Slack & Discord webhook alerts for crash diagnosis.
- Auto-merging approved low-risk patches directly into production.
Built With
- ai
- aws-bedrock
- cli
- devops
- github-api
- python
- rich
Log in or sign up for Devpost to join the conversation.