A Kubernetes Runbook Assistant built on Cloudflare Workers AI.
Describe a Kubernetes alert or incident in plain English and get step-by-step remediation guidance powered by Llama 3.3 70B.
- Cloudflare Workers — handles HTTP routing
- Durable Objects — stores conversation history per session
- Workers AI — runs Llama 3.3 70B inference
| Component | Technology |
|---|---|
| Runtime | Cloudflare Workers |
| Language | TypeScript |
| LLM | @cf/meta/llama-3.3-70b-instruct-fp8-fast |
| State | Durable Objects |
| Frontend | Vanilla HTML/CSS/JS served from Worker |
- Node.js 18+
- Cloudflare account
- Wrangler CLI
npm installnpm run dev -- --localnpm run deploy- Open the app URL in your browser
- Describe a Kubernetes incident in the chat input
- The assistant returns diagnosis, remediation steps, verification, and prevention advice
- Continue the conversation — the assistant remembers context
- Click + new chat to start a fresh session
- Pod stuck in CrashLoopBackOff in production namespace
- OOMKilled exit code 137 on Java application pods
- Pods stuck in Pending state — Insufficient cpu
- Node NotReady after cluster upgrade on EKS