airunway-aks-setup Skill for Claude: AI Runway on Azure Kubernetes

airunway-aks-setup walks Claude through setting up AI Runway on an Azure Kubernetes (AKS) cluster, from a bare cluster to a running model.

Installs: 366.4K; on skills.sh
Stars: 1.6K; on GitHub
Files: 11; besides SKILL.md
Updated: Oct 2026; last repo commit

airunway-aks-setup follows six steps in order: verify the cluster, install the AI Runway controller, assess GPUs, choose and install an inference provider (KAITO, Dynamo or KubeRay), deploy a first model and summarize. It reports each step as healthy or failed and asks you before any install or deployment. You can resume with skip-to-step N.

It assumes an AKS cluster exists. If not, it sends you to the azure-kubernetes skill first. It runs kubectl, make and curl against your cluster. GPU nodes are costly (the skill cites A100-80GB at about $3 to $5+ per hour), so it asks you to confirm costs first.

Good for

Platform engineers putting an LLM on an AKS cluster
Teams comparing KAITO, Dynamo and KubeRay as inference providers
Anyone resuming a half-finished AI Runway setup

Install airunway-aks-setup

Claude Code

Copy this prompt.

Install the airunway-aks-setup skill in Claude Code. Run this command:

npx skills add https://github.com/microsoft/azure-skills --skill airunway-aks-setup

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Set up AI Runway on AKS — from bare cluster to running model. Covers cluster verification, controller install, GPU assessment, provider setup, and first deployment. WHEN: "setup AI Runway", "onboard AKS cluster", "install AI Runway", "airunway setup", "deploy model to AKS", "GPU inference on AKS", "KAITO setup on AKS", "run LLM on AKS", "vLLM on AKS", "set up model serving on AKS", "AI Runway controller".

Sections in SKILL.md

Prerequisites
Quick Reference
When to Use This Skill
GPU setup versus Day-2 incidents
MCP Tools
Rules
Steps
Error Handling

Key facts

Author: Microsoft
License: MIT
Installs on skills.sh: 366,396 as of 2026-11-02
GitHub stars: 1,609 for the whole microsoft/azure-skills repo
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk pass
What is inside: 12 files: SKILL.md and 11 reference files (6 step files, GPU profiles, model sizing, PowerShell notes, troubleshooting, a day-2 GPU note); no scripts
Needs: An AKS cluster and kubeconfig access, and kubectl, make and curl
Last commit to the repo: 2026-10-30

What it includes

Checks context, nodes and GPU presence
Uses kubectl to confirm the cluster context, inventory nodes and detect GPUs.
Try: “Set up AI Runway on my AKS cluster.”
Runs code · 1. Cluster verification
Installs the CRDs and AI Runway controller
Deploys into the airunway-system namespace after your confirmation.
Runs code · 2. Controller installation
Detects GPU models and flags limits
Flags constraints such as T4 and V100 lacking bfloat16, which needs --dtype float16.
Runs code · 3. GPU assessment
Recommends and installs KAITO, Dynamo or KubeRay
Installs the inference provider you choose.
Runs code · 4. Provider setup
Picks a model, deploys it, waits for Ready
Creates a ModelDeployment and checks it is Ready.
Runs code · 5. First deployment
Recap, smoke test and next steps
Ends with a smoke test.
Runs code · 6. Summary
Resumes from a given step
Assumes earlier steps are done.
Read · skip-to-step N
Error handling and rollback procedures
Loaded when a step fails, for example a controller in CrashLoopBackOff.
Read · references/troubleshooting.md

From the official docs

Permissions

Medium risk

It installs software on a real cluster and can start costly GPU workloads. Check which cluster your kubectl context points to, and approve each step yourself.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It installs on whatever cluster kubectl points to
Step 1 checks the context. Confirm it is the cluster you mean before approving the controller install.
GPU deployments can cost real money
SKILL.md cites A100-80GB at roughly $3 to $5+ per hour. Set a budget before you approve a GPU node pool.
It runs make targets and curl
Review each command Claude proposes before you approve it.

Technical details

Talks to	Your Kubernetes cluster and the package or image sources it pulls from
Scopes	Runs kubectl, make and curl against the cluster your kubeconfig points to Installs CRDs, the AI Runway controller and an inference provider Deploys models onto GPU nodes, which cost money

Questions

What does airunway-aks-setup do?

It guides Claude through installing AI Runway on an existing AKS cluster and deploying a first AI model, in six confirmed steps.

How do I install airunway-aks-setup?

Use the install box at the top of this page. Microsoft's Azure skills plugin also bundles it, per the azure-skills README.

Do I need an AKS cluster first?

Yes. SKILL.md says to hand off to the azure-kubernetes skill to provision one if you do not have a cluster, with a GPU node pool unless CPU-only inference is fine.

What will it cost?

SKILL.md warns that GPU node pools are expensive, citing A100-80GB at about $3 to $5+ per hour. It tells Claude to confirm cost with you before provisioning GPUs.

Will it change my cluster without asking?

No. Rule 3 says to ask for confirmation before any install or deployment action. Review each command before you approve it.

Does it need an API key or MCP server?

SKILL.md says it uses no MCP tools. It works through kubectl and make with your existing cluster credentials.

Is airunway-aks-setup safe?

It installs controllers and deploys workloads on a real cluster. On skills.sh the audits read Gen Agent Trust Hub pass, Socket pass and Snyk pass. Try a test cluster first.

Information

updated Nov 2, 2026

Repository: microsoft/azure-skills
Folder: skills/airunway-aks-setup

microsoft/azure-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3

airunway-aks-setup SKILL.md · github.com
microsoft/azure-skills README · github.com
skills.sh listing and audits · skills.sh