Cloud Error Fixer Agent

Your personal AI Site Reliability Engineer. Paste any complex GCP or Kubernetes error log and get instant root cause analysis and actionable fixes.

Comment

Inspiration 💡

In the fast-paced world of Cloud Computing and DevOps, Site Reliability Engineers (SREs) and Cloud Architects spend countless hours deciphering cryptic error logs, stack traces, and Kubernetes crash loops. The inspiration for the Cloud Error Fixer Agent came from a simple question: What if we could democratize enterprise-grade cloud expertise? I wanted to create an AI companion that doesn't just say "here is the error," but acts as an Elite GCP SRE—providing instant root-cause analysis, actionable CLI commands, and long-term prevention strategies.

What it does 🛠️

Cloud Error Fixer is a web-based AI diagnostic tool. Developers can paste any raw stack trace, Google Cloud platform error, or Kubernetes (GKE) log into the portal. Within seconds, the AI agent breaks down the problem into:

  1. Root Cause Analysis: Explaining why it failed in plain English.
  2. Technical Deep Dive: Understanding the architectural breakdown.
  3. Actionable Fix: Providing the exact gcloud, kubectl, or Terraform commands to resolve the issue.
  4. Prevention Strategy: Best practices to ensure the error never happens again.

How we built it ⚙️

The backend is powered by the newly released google-genai SDK, specifically leveraging the formidable Gemini 2.5 Pro model. Drawing from my background as a prompt engineering specialist, I spent significant time carefully crafting and refining the core system instruction. By heavily engineering the persona constraint, I forced the model to operate strictly as a "Production-Grade Cloud Error Fixer Agent," which drastically reduced hallucinations and ensured the outputs were formatted for enterprise-scale environments. The frontend is built using Python and Streamlit, providing a sleek, dark-themed, and highly responsive user interface. Finally, it was deployed seamlessly to the Streamlit Community Cloud.

Challenges we ran into 🚧

Transitioning from local development to cloud deployment presented some hurdles. Initially, deploying via Google Cloud Run brought up unforeseen billing and IAM prerequisite challenges. Additionally, migrating from the deprecated google.generativeai library to the brand-new google-genai package required refactoring the code mid-development to ensure we were using the most up-to-date Google technologies.

Accomplishments that we're proud of 🏆

I am incredibly proud of the UI/UX experience and the accuracy of the model. Feeding it a highly complex, real-world Kubernetes Istio Webhook timeout error and watching it instantly generate the precise GKE Firewall rules and debugging steps was a "wow" moment. Building a fully functional, production-ready AI web app from scratch and deploying it globally is a massive win.

What we learned 🧠

This project was a deep dive into the practical applications of Large Language Models in technical support. I learned how to integrate the latest Gemini API, manage cloud secrets securely, and effectively use Streamlit to rapidly prototype and deploy AI interfaces.

What's next for Cloud Error Fixer Agent 🚀

In the future, I plan to add:

  • Direct GCP Integration: Allowing the agent to securely pull logs directly from Cloud Logging via OAuth.
  • Architecture Diagram Generation: Using generative AI to draw where the failure happened in the topology.
  • Slack/Discord Bot Integration: So DevOps teams can tag the bot directly in their incident response channels.

Built With

Share this project:

Updates