How anshchoudhary/quick-kube works
Main components
• User / CLI Layer
– src/main.py (entry point)
– scripts/setup.sh (environment bootstrap)
– examples/sample_apps/docker-compose.yml (input)
• AI Agent Layer (pipeline steps)
– Analysis Agent
– Conversion Agent (Kompose orchestration)
– Helm Agent (chart generation)
– Terraform Agent (GKE provisioning)
– Deployment Agent (final deploy + monitoring)
• Conversion & Generation Tools
– Kompose (Docker → Kubernetes manifests)
– Helm (chart templating & packaging)
– Terraform (GKE infra as code)
• Outputs / Artifacts
– Kubernetes manifests (output/kubernetes/)
– Helm charts (output/helm/quick-kube-app/)
– Terraform configs (templates/terraform or inline)
• Target Cloud Environment
– Google Kubernetes Engine (GKE)
– Kubernetes API (kubectl)
• External Services
– AI Providers (OpenAI GPT, Anthropic Claude, Google Gemini)
– Google Cloud SDK
Relationships and data flows
• User invokes CLI with a Docker-Compose file
└▶ Orchestrator (src/ai_agents/orchestrator.py)
• Orchestrator calls in sequence:
1. Analysis Agent → inspects containers, decides strategy
2. Conversion Agent → invokes Kompose → produces raw K8s manifests
3. Helm Agent → consumes manifests + templates → emits Helm chart
4. Terraform Agent → renders Terraform templates → provisions GKE cluster
5. Deployment Agent → applies Helm chart (helm install/upgrade) → monitors
• Each AI agent calls external AI APIs to generate or refine templates
• Terraform Agent applies to GCP via google_cloud SDK credentials
• Deployment Agent uses kubectl/Helm CLI to push to GKE
Architectural patterns
• Pipeline / Orchestration pattern (linear workflow)
• Agent-based modularity (each AI agent is a plug-in)
• Infrastructure as Code (Terraform, Helm)
• Template-driven code generation
Key technologies & frameworks
• Python 3.8+ (modules under src/, utils, ai_agents)
• Kompose CLI
• Helm
• Terraform
• Google Cloud SDK (gcloud, kubectl)
• AI Provider SDKs / HTTP APIs