Azure Kubernetes Service (AKS)

Azure Kubernetes Service (AKS) MCP Server

Azure Kubernetes Service (AKS) official MCP server.

What is an MCP Server?

MCP Info

Image Building Info

Available Tools (15)

Tools Details

Tool: az_advisor_recommendation

Retrieve and manage Azure Advisor recommendations for AKS clusters

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: az_aks_operations

Unified tool for managing Azure Kubernetes Service (AKS) clusters and related operations.

Supported operations:

• Cluster: show, list, get-versions, check-network
• Nodepool: nodepool-list, nodepool-show
• Account: account-list

Examples:

• Show cluster: operation="show", args="--name myCluster --resource-group myRG"
• List nodepools: operation="nodepool-list", args="--cluster-name myCluster --resource-group myRG"

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: az_compute_operations

Unified tool for managing Azure Virtual Machines (VMs) and Virtual Machine Scale Sets (VMSS) using Azure CLI.

IMPORTANT: VM/VMSS resources are managed by AKS. Write operations should be used carefully and only for debugging purposes.

Use resource_type="vm" for single virtual machines or resource_type="vmss" for virtual machine scale sets.

Available operation values:

• show: Get details of a VM/VMSS
• list: List VMs/VMSS in subscription or resource group
• get-instance-view: Get runtime status

EXAMPLES: List VMSS: operation="list", resource_type="vmss", args="--resource-group myRG" Show VMSS: operation="show", resource_type="vmss", args="--name myVMSS --resource-group myRG" List VMs: operation="list", resource_type="vm", args="--resource-group myRG"

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: az_fleet

Run Azure Kubernetes Service Fleet management commands.

Available operations and resources:

• fleet: list, show, create, update, delete, get-credentials
• member: list, show, create, update, delete
• updaterun: list, show, create, start, stop, delete
• updatestrategy: list, show, create, delete
• clusterresourceplacement: list, show, get, create, delete (Kubernetes CRD operations)

Examples:

• List fleets: operation='list', resource='fleet', args='--resource-group myRG'
• Show fleet: operation='show', resource='fleet', args='--name myFleet --resource-group myRG'
• Get fleet credentials: operation='get-credentials', resource='fleet', args='--name myFleet --resource-group myRG'
• Create member: operation='create', resource='member', args='--name myMember --fleet-name myFleet --resource-group myRG --member-cluster-id /subscriptions/.../myCluster'
• Create clusterresourceplacement: operation='create', resource='clusterresourceplacement', args='--name nginx --selector app=nginx --policy PickAll'
• List clusterresourceplacements: operation='list', resource='clusterresourceplacement', args=''

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: az_monitoring

Unified tool for Azure monitoring and diagnostics operations for AKS clusters.

Supported Operations:

1. Metrics - Query Azure Monitor metrics for AKS clusters and nodes

   • list: Get metric values for specific metrics
   • list-definitions: Get available metrics for a resource
   • list-namespaces: Get metric namespaces for a resource

   Use for: CPU usage, memory consumption, network traffic, pod counts, node health Required parameters: resource (Azure resource ID) Additional for 'list': metrics (metric names) Optional: aggregation, start-time, end-time, interval, filter

2. Resource Health - Get Azure Resource Health events for AKS clusters Use for: Cluster availability issues, platform problems, service health events Required parameters: subscription_id, resource_group, cluster_name, start_time Optional: end_time, status (Available, Unavailable, Degraded, Unknown)

3. Application Insights - Execute KQL queries against Application Insights telemetry Use for: Application performance monitoring, custom telemetry analysis, trace correlation Required parameters: subscription_id, resource_group, app_insights_name, query Optional: start_time + end_time OR timespan (not both)

4. Diagnostics - Check AKS cluster diagnostic settings configuration Use for: Verify logging is enabled, check log retention, validate diagnostic configuration Required parameters: subscription_id, resource_group, cluster_name

5. Control Plane Logs - Query AKS control plane logs Supported log categories:

   • kube-apiserver
   • kube-audit
   • kube-audit-admin
   • kube-controller-manager
   • kube-scheduler
   • cluster-autoscaler
   • cloud-controller-manager
   • guard (for authentication/authorization issues)
   • csi-azuredisk-controller
   • csi-azurefile-controller
   • csi-snapshot-controller
   • fleet-member-agent
   • fleet-member-net-controller-manager
   • fleet-mcs-controller-manager PLEASE NOTE: you need to check if the category is enabled in your cluster's diagnostic settings by using the diagnostics tool.

Use This Tool When You Need To:

• Monitor cluster or other azure resource performance and usage (use metrics)
• Check cluster availability and platform health (use resource_health)
• Analyze application telemetry and performance (use app_insights)
• Verify diagnostic logging configuration (use diagnostics)
• Debug Kubernetes API server issues (use control_plane_logs with kube-apiserver)
• Investigate authentication/authorization problems (use control_plane_logs with kube-audit, guard)
• Troubleshoot pod scheduling issues (use control_plane_logs with kube-scheduler)
• Check storage-related problems (use control_plane_logs with csi-azuredisk-controller, csi-azurefile-controller)
• Analyze cluster scaling behavior (use control_plane_logs with cluster-autoscaler)
• Review security audit events (use control_plane_logs with kube-audit, kube-audit-admin)

Examples:

metrics:

• Get CPU usage: operation="metrics", query_type="list", parameters="{"resource":"/subscriptions/sub-id/resourceGroups/rg/providers/Microsoft.ContainerService/managedClusters/cluster", "metrics":"node_cpu_usage_percentage", "aggregation":"Average", "start-time":"<start-time>", "end-time":"<end-time>"}"
• List available metrics: operation="metrics", query_type="list-definitions", parameters="{"resource":"/subscriptions/sub-id/resourceGroups/rg/providers/Microsoft.ContainerService/managedClusters/cluster"}"

resource_health:

• Check recent cluster health: operation="resource_health", subscription_id="<subscription-id>", resource_group="<resource-group>", cluster_name="<cluster-name>", parameters="{"start_time":"<start-time>"}"

app_insights:

• Query request telemetry: operation="app_insights", subscription_id="<subscription-id>", resource_group="<resource-group>", parameters="{"app_insights_name":"myapp-insights", "query":"requests | where timestamp > ago(1h) | summarize count() by bin(timestamp, 5m)"}"
• Analyze exceptions: operation="app_insights", subscription_id="<subscription-id>", resource_group="<resource-group>", parameters="{"app_insights_name":"myapp-insights", "query":"exceptions | where timestamp > ago(24h) | summarize count() by type, bin(timestamp, 1h)"}"
• Performance with timespan: operation="app_insights", subscription_id="<subscription-id>", resource_group="<resource-group>", parameters="{"app_insights_name":"myapp-insights", "query":"performanceCounters | where category == 'Processor' | summarize avg(value) by bin(timestamp, 5m)", "timespan":"PT1H"}"

diagnostics:

• Verify diagnostic settings: operation="diagnostics", subscription_id="<subscription-id>", resource_group="<resource-group>", cluster_name="<cluster-name>", parameters="{}"

control_plane_logs:

• Query API server logs: operation="control_plane_logs", subscription_id="<subscription-id>", resource_group="<resource-group>", cluster_name="<cluster-name>", parameters="{"log_category":"kube-apiserver", "start_time":"<start-time>", "end_time":"<end-time>", "max_records":"50"}"
• Debug authentication issues: operation="control_plane_logs", subscription_id="<subscription-id>", resource_group="<resource-group>", cluster_name="<cluster-name>", parameters="{"log_category":"guard", "start_time":"<start-time>", "end_time":"<end-time>", "max_records":"100"}"
• Analyze audit events: operation="control_plane_logs", subscription_id="<subscription-id>", resource_group="<resource-group>", cluster_name="<cluster-name>", parameters="{"log_category":"kube-audit", "log_level":"error", "start_time":"<start-time>", "end_time":"<end-time>", "max_records":"50"}"

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: az_network_resources

Unified tool for getting Azure network resource information used by AKS clusters.

Supported resource types:

• all: Get information about all network resources
• vnet: Get Virtual Network information
• nsg: Get Network Security Group information
• route_table: Get Route Table information
• subnet: Get Subnet information
• load_balancer: Get Load Balancer information
• private_endpoint: Get Private Endpoint information (private clusters only)

Examples:

• Get all network resources: resource_type="all"
• Get VNet info: resource_type="vnet"
• Get NSG info: resource_type="nsg"

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: get_aks_vmss_info

Get detailed VMSS configuration for a specific node pool or all node pools in the AKS cluster (provides low-level VMSS settings not available in az aks nodepool show). Leave node_pool_name empty to get info for all node pools.

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: inspektor_gadget_observability

Real-time observability tool for Azure Kubernetes Service (AKS) clusters, allowing users to manage gadgets for monitoring and debugging

Apart from 'action' param:

It supports 'action_params' (type=object) to specify parameters for the action.Available params are: gadget_name, duration, gadget_id, chart_version. Available Gadget names are: observe_dns, observe_tcp, observe_file_open, observe_process_execution, observe_signal, observe_system_calls, top_file, top_tcp, tcpdump. Example: {'action': 'run', 'action_params': {'gadget_name': 'observe_dns', 'duration': 10}}

It supports 'filter_params' (type=object) to filter the data captured by the gadget. Available params are: namespace, pod, container, selector,observe_signal.signal, observe_dns.nameserver, observe_dns.minimum_latency, observe_dns.response_code, observe_dns.unsuccessful_only, observe_tcp.destination_port, observe_tcp.event_type, observe_file_open.path, observe_dns.name, observe_file_open.unsuccessful_only, observe_process_execution.command, observe_system_calls.syscall, top_tcp.max_entries, tcpdump.pcap-filter, observe_tcp.unsuccessful_only, observe_tcp.source_port, top_file.max_entries. Example: {'action': 'run', 'filter_params': {'namespace': 'default', 'selector': 'app=myapp', 'observe_dns.unsuccessful_only': true}}

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: kubectl_cluster

Get information about the Kubernetes cluster and API.

Available operations:

• cluster-info: Display cluster information
• api-resources: Print supported API resources
• api-versions: Print supported API versions
• explain: Get documentation for a resource

Examples:

• Cluster info: operation='cluster-info', resource='', args=''
• Cluster info dump: operation='cluster-info', resource='dump', args=''
• List resources: operation='api-resources', resource='', args=''
• Namespaced resources: operation='api-resources', resource='', args='--namespaced=true'
• Non-namespaced resources: operation='api-resources', resource='', args='--namespaced=false'
• Resources by group: operation='api-resources', resource='', args='--api-group=rbac.authorization.k8s.io'
• API versions: operation='api-versions', resource='', args=''
• Explain pod: operation='explain', resource='pods', args=''
• Explain field: operation='explain', resource='pods.spec.containers', args=''
• Explain with version: operation='explain', resource='deployments', args='--api-version=apps/v1'

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: kubectl_config

Work with Kubernetes configurations (read-only).

Available operations:

• diff: Diff the live version against what would be applied
• auth: Inspect authorization (can-i)
• config: View kubectl configuration contexts (read-only)

Config operations (read-only):

• current-context: Display the current context
• get-contexts: Describe one or many contexts

Examples:

• Diff config: operation='diff', resource='', args='-f pod.json'
• Check auth: operation='auth', resource='can-i', args='create pods --all-namespaces'
• Get current context: operation='config', resource='current-context', args=''
• List contexts: operation='config', resource='get-contexts', args=''

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: kubectl_diagnostics

Diagnose and debug Kubernetes resources.

Available operations:

• logs: Print logs for a container in a pod
• events: Display events
• top: Display resource usage (CPU/Memory)
• exec: Execute a command in a container
• cp: Copy files to/from containers

Examples:

• Logs for default container: operation='logs', resource='', args='nginx'
• Logs for specific container: operation='logs', resource='', args='nginx -c ruby-container'
• Logs with selector: operation='logs', resource='', args='-l app=nginx --all-containers=true'
• Get events: operation='events', resource='', args='--all-namespaces'
• Get events namespace: operation='events', resource='', args='-n default'
• Top pods: operation='top', resource='pod', args=''
• Top nodes: operation='top', resource='node', args=''
• Top with containers: operation='top', resource='pod', args='POD_NAME --containers'
• Exec command: operation='exec', resource='', args='mypod -n NAMESPACE -- date'
• Copy to pod: operation='cp', resource='', args='/tmp/foo_dir some-pod:/tmp/bar_dir'
• Copy from pod: operation='cp', resource='', args='some-namespace/some-pod:/tmp/foo /tmp/bar'
• Copy with container: operation='cp', resource='', args='/tmp/foo some-pod:/tmp/bar -c specific-container'

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: kubectl_resources

View Kubernetes resources with read-only operations.

Available operations:

• get: Display one or many resources
• describe: Show detailed information about resources

Common resources: pods, deployments, services, configmaps, secrets, namespaces, etc.

Examples:

• Get pods: operation='get', resource='pods', args='-n default'
• Get specific pod: operation='get', resource='pods', args='nginx-pod -n default'
• Get with selector: operation='get', resource='pods', args='-l app=nginx'
• Get all namespaces: operation='get', resource='pods', args='--all-namespaces'
• Describe deployment: operation='describe', resource='deployment', args='myapp -n production'
• Describe all pods: operation='describe', resource='pods', args=''
• Describe with selector: operation='describe', resource='pods', args='-l name=myLabel'

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: list_detectors

List all available AKS cluster detectors

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: run_detector

Run a specific AKS detector

This tool may perform destructive updates.

This tool interacts with external entities.

Tool: run_detectors_by_category

Run all detectors in a specific category

This tool may perform destructive updates.

This tool interacts with external entities.

Use this MCP Server

{
  "mcpServers": {
    "aks": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-v",
        "/local-directory:/local-directory",
        "mcp/aks",
        "--transport=stdio",
        "--access-level={{aks.access_level}}",
        "--allow-namespaces={{aks.allow_namespaces}}",
        "--additional-tools={{aks.additional_tools}}"
      ]
    }
  }
}

Why is it safer to run MCP Servers with Docker?