[WIP] Kagent for Kueue

Makefile

@@ -291,6 +291,8 @@ helm-agents:
 	helm package -d $(HELM_DIST_FOLDER) helm/agents/istio
 	VERSION=$(VERSION) envsubst < helm/agents/promql/Chart-template.yaml > helm/agents/promql/Chart.yaml
 	helm package -d $(HELM_DIST_FOLDER) helm/agents/promql
+	VERSION=$(VERSION) envsubst < helm/agents/kueue/Chart-template.yaml > helm/agents/kueue/Chart.yaml
+	helm package -d $(HELM_DIST_FOLDER) helm/agents/kueue
 	VERSION=$(VERSION) envsubst < helm/agents/observability/Chart-template.yaml > helm/agents/observability/Chart.yaml
 	helm package -d $(HELM_DIST_FOLDER) helm/agents/observability
 	VERSION=$(VERSION) envsubst < helm/agents/helm/Chart-template.yaml > helm/agents/helm/Chart.yaml
@@ -376,6 +378,7 @@ helm-publish: helm-version
 	helm push ./$(HELM_DIST_FOLDER)/kagent-$(VERSION).tgz $(HELM_REPO)/kagent/helm
 	helm push ./$(HELM_DIST_FOLDER)/helm-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents
 	helm push ./$(HELM_DIST_FOLDER)/istio-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents
+    helm push ./$(HELM_DIST_FOLDER)/kueue-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents
 	helm push ./$(HELM_DIST_FOLDER)/promql-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents
 	helm push ./$(HELM_DIST_FOLDER)/observability-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents
 	helm push ./$(HELM_DIST_FOLDER)/argo-rollouts-agent-$(VERSION).tgz $(HELM_REPO)/kagent/agents

helm/agents/kueue/Chart-template.yaml

@@ -0,0 +1,5 @@
+apiVersion: v2
+name: kueue-agent
+description: A Kueue Agent for kagent
+type: application
+version: ${VERSION}

helm/agents/kueue/templates/agent.yaml

@@ -0,0 +1,67 @@
+apiVersion: kagent.dev/v1alpha2
+kind: Agent
+metadata:
+  name: kueue-conversion-agent
+  namespace: {{ include "kagent.namespace" . }}
+  labels:
+    {{- include "kagent.labels" . | nindent 4 }}
+spec:
+  description: The Kueue AI Agent specializes in assisting users to understand their Kueue configuration, including Queues, Priorities and Workloads.
+  type: Declarative
+  declarative:
+    systemMessage: |-
+      You are an Kubernetes Kueue specialist focused jobs schedulling. You
+      are only responsible for helping the user to understand their Kueue configuration, and resources, including Queues and Priorities.
+
+      Your key responsibility is assisting users with scheduling their jobs based on existing queues and priorities.
+
+
+      To achieve this use the k8s tools by filtering ClusterQueues, Queues, and PriorityClasses.
+      ```
+      admissionchecks.kueue.x-k8s.io              
+      clusterqueues.kueue.x-k8s.io                
+      cohorts.kueue.x-k8s.io                      
+      localqueues.kueue.x-k8s.io                  
+      multikueueclusters.kueue.x-k8s.io           
+      multikueueconfigs.kueue.x-k8s.io            
+      provisioningrequestconfigs.kueue.x-k8s.io   
+      resourceflavors.kueue.x-k8s.io              
+      topologies.kueue.x-k8s.io                   
+      workloadpriorityclasses.kueue.x-k8s.io      
+      workloads.kueue.x-k8s.io  
+      ```
+
+      When providing information about Kueue concepts, always refer to the official Kueue documentation at https://kueue.sigs.k8s.io/ for accurate and up-to-date information.
+      As part of your job, you can teach the user how to optimally use Kueue for their job scheduling needs.
+
+    modelConfig: {{ .Values.modelConfigRef | default (printf "%s" (include "kagent.defaultModelConfigName" .)) }}
+    tools:
+    - type: McpServer
+      mcpServer:
+        name: kagent-tool-server
+        kind: RemoteMCPServer
+        apiGroup: kagent.dev
+        toolNames:
+        - k8s_get_resources
+        - k8s_get_resource_yaml
+        - k8s_create_resource
+        - k8s_apply_manifest
+        - k8s_describe_resource
+    a2aConfig:
+      skills:
+        - id: queues
+          name: Provide information about Kueue Queues and Workfloads
+          description: Check the current cluster to see which resources are defined and check their status.
+          tags:
+            - queues
+            - workloads
+            - kueue
+            - scheduling
+            - resources
+          examples:
+            - "Show me the list of Kueue Queues and their statuses in my cluster."
+            - "Show me the Kueue Workloads and their priorities."
+    deployment:
+      resources:
+        {{- toYaml .Values.resources | nindent 8 }}
+

helm/agents/kueue/templates/rbac.yaml

@@ -0,0 +1,32 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: {{ include "kagent.fullname" . }}-kueue-role
+  labels:
+    {{- include "kagent.labels" . | nindent 4 }}
+rules:
+- apiGroups:
+    - kueue.x-k8s.io
+  resources:
+    - clusterqueues
+    - localqueues
+    - resourceflavors
+    - workloads
+    - workloadpriorityclasses
+  verbs:
+    - "*"
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: {{ include "kagent.fullname" . }}-kueue-rolebinding
+  labels:
+    {{- include "kagent.labels" . | nindent 4 }}
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: {{ include "kagent.fullname" . }}-kueue-role
+subjects:
+  - kind: ServiceAccount
+    name: {{ include "kagent.fullname" . }}
+    namespace: {{ include "kagent.namespace" . }}

helm/agents/kueue/values.yaml

@@ -0,0 +1,9 @@
+modelConfigRef: ""
+
+resources:
+  requests:
+    cpu: 100m
+    memory: 256Mi
+  limits:
+    cpu: 1000m
+    memory: 1Gi

helm/kagent/Chart-template.yaml

@@ -40,6 +40,10 @@ dependencies:
     version: ${VERSION}
     repository: file://../agents/observability
     condition: agents.observability-agent.enabled
+  - name: kueue-agent
+    version: ${VERSION}
+    repository: file://../agents/kueue
+    condition: agents.kueue-agent.enabled
   - name: argo-rollouts-agent
     version: ${VERSION}
     repository: file://../agents/argo-rollouts

helm/kagent/values.yaml

@@ -267,6 +267,15 @@ agents:
       limits:
         cpu: 1000m
         memory: 1Gi
+  kueue-agent:
+    enabled: true
+    resources:
+      requests:
+        cpu: 100m
+        memory: 256Mi
+      limits:
+        cpu: 1000m
+        memory: 1Gi
   observability-agent:
     enabled: true
     resources: