Add prerequisites to lab documentation

docs/labs/README.md

@@ -2,6 +2,16 @@
 
 Hands-on labs for trying AKS Flex Node in end-to-end Azure scenarios.
 
+## Common Prerequisites
+
+Before starting a lab, prepare:
+
+- An Azure subscription where you can create resource groups, VNets, VMs, AKS clusters, and private DNS links.
+- Azure CLI logged in to the target subscription.
+- `kubectl`, Helm, `curl`, and SSH/SCP tooling on the workstation or admin VM that will run the lab commands.
+- Non-overlapping CIDR ranges for the AKS VNet, Flex VM VNet, pod networks, service CIDR, and any connected networks.
+- Network access from the command runner and Flex VM to the AKS API server. For private AKS labs, run `kubectl`, Helm, and bootstrap config commands from a machine that can resolve and reach the private API endpoint.
+
 ## Available Labs
 
 - [GPU Flex Node setup](gpu-node-setup.md) - GPU host image and driver contract, cluster GPU stack, validation, and troubleshooting.

docs/labs/aks-private-cluster-cilium.md

@@ -6,6 +6,15 @@ The validated setup uses AKS private cluster mode with `--network-plugin none` a
 
 For Cilium concepts and operations, see the [Cilium documentation](https://docs.cilium.io/).
 
+## Prerequisites
+
+- An Azure subscription where you can create resource groups, VNets, VMs, a private AKS cluster, private DNS links, and the bootstrap RBAC needed by AKS Flex Node.
+- Azure CLI logged in to the target subscription.
+- `kubectl`, Helm, `curl`, `python3`, and SSH/SCP tooling on the workstation or admin VM that will run the lab commands.
+- A command runner that can resolve and reach the private AKS API endpoint. If your workstation cannot, use the admin VM described below.
+- Non-overlapping CIDR ranges for the AKS VNet, Flex VM VNet, Cilium pod CIDR, AKS service CIDR, and any connected networks.
+- A Flex VM image with Ubuntu 24.04 and sudo access.
+
 ## What Is Cilium?
 
 Cilium is an eBPF-based networking, security, and observability layer for Kubernetes. In this setup, Cilium provides CNI functionality for both the AKS system node and the Flex node.

docs/labs/aks-private-cluster-unbounded-net.md

@@ -6,6 +6,15 @@ The validated setup uses AKS private cluster mode with `--network-plugin none` a
 
 For unbounded-net concepts, custom resources, and operations, see the [Unbounded networking documentation](https://unbounded-cloud.io/concepts/networking/) and [unbounded-net operations guide](https://unbounded-cloud.io/reference/networking/operations/).
 
+## Prerequisites
+
+- An Azure subscription where you can create resource groups, VNets, VMs, a private AKS cluster, private DNS links, and the bootstrap RBAC needed by AKS Flex Node.
+- Azure CLI logged in to the target subscription.
+- `kubectl`, `curl`, `git`, `make`, `python3`, and SSH/SCP tooling on the workstation or admin VM that will run the lab commands.
+- A command runner that can resolve and reach the private AKS API endpoint. If your workstation cannot, use the admin VM described below.
+- Non-overlapping CIDR ranges for the AKS VNet, Flex VM VNet, AKS pod CIDR, Flex pod CIDR, AKS service CIDR, and any connected networks.
+- A Flex VM image with Ubuntu 24.04 and sudo access.
+
 ## What Is Unbounded-Net?
 
 `unbounded-net` is the networking layer from [Project Unbounded](https://unbounded-cloud.io/). It provides CNI functionality and multi-site pod networking for Kubernetes clusters whose nodes may live in different networks, regions, or clouds.

docs/labs/gpu-node-setup.md

@@ -15,10 +15,12 @@ Plan for both before you start.
 
 ## Before you begin
 
-- An AKS cluster with `kubectl` admin access.
-- A GPU host with root access and outbound reach to the AKS API server.
+- An Azure subscription and AKS cluster with `kubectl` admin access.
+- Azure CLI logged in to the target subscription.
+- `kubectl`, Helm, `curl`, and SSH/SCP tooling on your workstation.
+- A GPU host with root or sudo access and outbound reach to the AKS API server.
 - A GPU host image that already includes the NVIDIA driver.
-- Helm installed on your workstation to install the cluster GPU stack.
+- Non-overlapping network ranges for the AKS cluster, host network, pods, services, and any connected networks.
 
 ## Driver and image contract