Skip to main content

DevOps AI Toolkit Stack

Deploy the complete DevOps AI Toolkit stack with a single Helm command.

Overview​

The dot-ai-stack umbrella chart installs all DevOps AI Toolkit components together with a single command:

  • DevOps AI Toolkit - MCP server for AI-powered Kubernetes operations
  • DevOps AI Toolkit Controller - Kubernetes controller for intelligent resource management and autonomous operations
  • DevOps AI Toolkit Web UI - Web interface for visual cluster management

Note: This guide covers Kubernetes deployment using the umbrella Helm chart. For other installation options (Docker, NPX, individual charts, etc.), see devopstoolkit.ai.

Prerequisites​

  • Kubernetes cluster with an ingress controller
  • Helm 3.x installed
  • kubectl configured with cluster access
  • AI API keys for AI-powered features (Anthropic and/or OpenAI)

Step 1: Create a Local Cluster (Optional)​

Skip this step if you already have a Kubernetes cluster with an ingress controller.

Create a Kind cluster with ingress port mappings:

kind create cluster --name dot-ai-stack --config - <<EOF
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
- role: control-plane
  extraPortMappings:
  - containerPort: 80
    hostPort: 80
    protocol: TCP
  - containerPort: 443
    hostPort: 443
    protocol: TCP
EOF

You should see output ending with:

Set kubectl context to "kind-dot-ai-stack"

Install the nginx ingress controller for Kind:

kubectl apply \
    --filename https://raw.githubusercontent.com/kubernetes/ingress-nginx/main/deploy/static/provider/kind/deploy.yaml

Wait for the ingress controller to be ready:

kubectl wait --namespace ingress-nginx \
    --for=condition=ready pod \
    --selector=app.kubernetes.io/component=controller \
    --timeout=90s

Step 2: Set Environment Variables​

Set your API keys for AI-powered features:

export ANTHROPIC_API_KEY="your-anthropic-api-key"

export OPENAI_API_KEY="your-openai-api-key"

Note: Multiple AI providers are supported. See AI Model Configuration for all options including Google Gemini, AWS Bedrock, Azure OpenAI, and others.

Generate a random authentication token for the MCP server:

export DOT_AI_AUTH_TOKEN=$(openssl rand -base64 32)

Step 3: Install the Stack​

Install the complete dot-ai stack with a single Helm command:

helm upgrade --install dot-ai-stack \
    oci://ghcr.io/vfarcic/dot-ai-stack/charts/dot-ai-stack \
    --namespace dot-ai --create-namespace \
    --set dot-ai.secrets.anthropic.apiKey=$ANTHROPIC_API_KEY \
    --set dot-ai.secrets.openai.apiKey=$OPENAI_API_KEY \
    --set dot-ai.secrets.auth.token=$DOT_AI_AUTH_TOKEN \
    --set dot-ai.ingress.enabled=true \
    --set dot-ai.ingress.className=nginx \
    --set dot-ai.ingress.host=dot-ai.127.0.0.1.nip.io \
    --set dot-ai-ui.ingress.enabled=true \
    --set dot-ai-ui.ingress.host=dot-ai-ui.127.0.0.1.nip.io \
    --wait

Note: Replace the ingress hosts with your actual domain names for production deployments.

This installs:

  • dot-ai - MCP server with ingress at dot-ai.127.0.0.1.nip.io
  • dot-ai-controller - Kubernetes controller for autonomous operations
  • dot-ai-ui - Web interface at dot-ai-ui.127.0.0.1.nip.io
  • Qdrant - Vector database for pattern and policy storage
  • ResourceSyncConfig - Enables resource discovery
  • CapabilityScanConfig - Enables cluster capability scanning

Step 4: Verify Installation​

Check that all pods are running:

kubectl get pods --namespace dot-ai

You should see all pods in Running status:

NAME                                        READY   STATUS    RESTARTS   AGE
dot-ai-577db5b4fc-j8kgf                     1/1     Running   0          50s
dot-ai-controller-manager-c898b5697-dqk2m   1/1     Running   0          50s
dot-ai-stack-qdrant-0                       1/1     Running   0          50s
dot-ai-ui-69d586db8b-ccqrm                  1/1     Running   0          50s

Test the MCP server health:

curl -H "Authorization: Bearer $DOT_AI_AUTH_TOKEN" \
    http://dot-ai.127.0.0.1.nip.io/healthz

Expected output:

{"status":"ok"}

Step 5: Configure MCP Client​

Create the MCP client configuration file:

cat > .mcp.json << EOF
{
  "mcpServers": {
    "dot-ai": {
      "type": "http",
      "url": "http://dot-ai.127.0.0.1.nip.io",
      "headers": {
        "Authorization": "Bearer $DOT_AI_AUTH_TOKEN"
      }
    }
  }
}
EOF

Note: This example creates .mcp.json in the current directory for Claude Code. Other MCP-enabled agents may expect the configuration in a different location (e.g., ~/.config/ or within the agent's settings). Consult your agent's documentation for the correct path.

Step 6: Start Using​

Launch your MCP-enabled AI tool:

claude

Note: The claude command is for Claude Code. If you're using a different MCP-enabled agent (e.g., Cursor, VS Code with MCP extension), launch it according to its documentation.

Try these example prompts:

What You WantExample Prompt
Check system status"Show dot-ai status"
Query cluster"What pods are running in the dot-ai namespace?"
List capabilities"List all capabilities"
Deploy an app"I want to deploy a web application"
Fix issues"Something is wrong with my database"

Configuration​

Override any component value by prefixing with the chart name:

--set dot-ai.resources.limits.memory=4Gi
--set dot-ai-controller.resources.limits.memory=1Gi
--set dot-ai-ui.ingress.host=ui.example.com

For available options, see each component's documentation:

Next Steps​

Cleanup​

To remove the stack:

helm uninstall dot-ai-stack --namespace dot-ai

kubectl delete namespace dot-ai

To delete the Kind cluster:

kind delete cluster --name dot-ai-stack