Skip to main content

Installing

Prerequisites​

Before starting, ensure you have the following tools installed:

Installation Steps​

  1. clone the repositories and create the kind cluster.
git clone https://github.com/rogan297/ai-learning-platform.git
cd ai-learning-platform
  1. Create the kind cluster.
kind create cluster
  1. Install the necessary CRDs for the Gateway API and the Agent components to ensure the cluster recognizes these resource types.
# Install Agent Gateway CRDs
helm upgrade -i --create-namespace \
  --namespace agentgateway-system \
  --version v2.2.1 agentgateway-crds \
  oci://ghcr.io/kgateway-dev/charts/agentgateway-crds

# Install KAgent CRDs
helm install kagent-crds oci://ghcr.io/kagent-dev/kagent/helm/kagent-crds \
  --namespace kagent \
  --create-namespace

# Install Standard Kubernetes Gateway API
kubectl apply -f https://github.com/kubernetes-sigs/gateway-api/releases/download/v1.4.0/standard-install.yaml

  1. Deploy the supporting services: Identity Management (Keycloak), Certificate Management, and the Database.
# Identity Provider
helm upgrade --install keycloak ./keycloak -n keycloak --create-namespace

# NOTE: You must manually create users and the Keycloak Realm.
# Default configuration: 
# keycloak realm: agent-realm
# username: agent
# password: 1234
# client_id: agent-client
# audience: agent-client
# Access the UI via port-forward:
kubectl port-forward svc/keycloak -n keycloak 8080:8080

# Database
helm upgrade --install postgres ./postgresql -n database --create-namespace
  1. Deploy the LGTM stack (Loki, Grafana, Tempo) and OpenTelemetry for monitoring and auditing.
# OpenTelemetry Collector
helm upgrade --install opentelemetry-collector-audit ./opentelemetry-collector \
  -n telemetry --create-namespace \
  -f opentelemetry-collector/otel-collector-audit-values.yaml

# Logs, Traces, and Dashboards
helm upgrade --install loki ./loki -n telemetry --create-namespace -f loki/loki-values.yaml
helm upgrade --install tempo ./tempo -n telemetry --create-namespace -f tempo/tempo-values.yaml
helm upgrade --install grafana ./grafana -n telemetry --create-namespace \
  -f grafana/grafana-values.yaml \
  --set assertNoLeakedSecrets=false
  1. Finally, deploy the gateway and the custom learning agents.
# Install Agent Gateway
helm upgrade --install agentgateway ./agentgateway/agentgateway-install \
  -n agentgateway-system \
  --create-namespace \
  --set auth.keycloak.realm="agent-realm" \
  --set auth.jwt.audience="agent-client" \
  --set controlPlane.serviceName="mcp-agent-control-plane"

# Before upload the docker images, you have to set the api keys and providers to:
nano agents/orchestrator/agent.yaml
nano mcp-agent-control-plane/agent-template.yaml 


# Build and Load Local Image (for Kind environments)
docker compose build
kind load docker-image agent/custom/orchestrator:0.1.0 agent/custom/template-agent:0.1.0 mcp-agent-control-plane:0.1.0

# Install KAgent
helm upgrade --install kagent ./kagent/kagent-install \
  -n kagent \
  --create-namespace
  
# Monitor the MCP server status
kubectl get pod -n kagent -w

# Create the agent
kubectl apply -f agents/orchestrator/agent.yaml
  1. Once the deployment is complete, you can verify the status of the components and start using the platform.
kubectl get pods -A

# Grafana Dashboards:
kubectl port-forward svc/grafana -n telemetry 3000:80

# Kagent User Interfaz
kubectl port-forward -n kagent service/kagent-ui 8082:8080