Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
85 changes: 85 additions & 0 deletions website/blog/2026-03-26-agent-skills-for-aks/index.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,85 @@
---
title: "Turn your agents into AKS experts: Agent Skills for AKS"
date: 2026-03-26
description: "Announcing agent skills for AKS, easily pluggable into any agent to enhance them with built-in AKS expertise, safer workflows, and reduced developer overhead."
authors: [julia-yin]
tags: [agent, skills, mcp, developer]
draft: true
Copy link

Copilot AI Mar 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

draft: true keeps the post out of production builds. The PR description reads like a publication/announcement; either set draft: false before merging or clarify that this PR is intentionally landing as a draft-only post.

Copilot uses AI. Check for mistakes.
---

We're excited to announce **agent skills for AKS**, bringing AKS best practices and troubleshooting guidance directly into **any AI agent** of your choosing. The AKS-specific skills are now available as part of the **GitHub Copilot for Azure** extension, directly available through VS Code, Copilot CLI, and Claude and pluggable into any compatible agent.

For AKS, skills matter because AI agents already carry a good baseline of Kubernetes and AKS knowledge, but that knowledge is only as current as their training data and varies across models. What agents typically lack is the prescriptive, up-to-date guidance specific to AKS: what the AKS engineering team recommends today, the exact workflows and tools for real troubleshooting scenarios, and the guardrails that keep changes safe in production. Today, we're releasing the first set of skills covering the two areas teams ask for help with most: best practices for AKS clusters and scenario-based troubleshooting.

<!-- truncate -->

Copy link

Copilot AI Mar 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Per the blog post content guidelines, a hero image (typically ./hero-image.png) should be placed immediately after <!-- truncate -->. This post currently has no hero image and the folder contains only index.md, so it won’t meet the expected post structure.

Suggested change
![Agent skills for AKS hero diagram](./hero-image.png)

Copilot uses AI. Check for mistakes.
## What are agent skills?

Agent skills are an open standard pioneered by Anthropic for enhancing AI agents with domain-specific expertise in a token-efficient way. You install a skill once, and any compatible agent (such as GitHub Copilot, Claude, Gemini, or others) picks it up automatically for relevant prompts. They provide essential expertise to your agents while staying context efficient, as each skill loads only when your prompt is relevant to the skill's content:

- If you're not asking about AKS, the skill **stays out of the way** and doesn't add to your token usage.
- When you do ask an AKS-related question, the skill **activates automatically** and brings in the right guidance, commands, and context.

## Skills which are currently available

The first release includes the following skills:

1. A high level **AKS best practices** skill:
1. Sub-skills for **AKS troubleshooting**:
Comment on lines +27 to +28
Copy link

Copilot AI Mar 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These numbered list entries are incomplete (they end with : but don’t list the actual skill names/links/details). Also, “high level” should be hyphenated as “high-level” when used as an adjective.

Suggested change
1. A high level **AKS best practices** skill:
1. Sub-skills for **AKS troubleshooting**:
1. A high-level **AKS best practices** skill that covers cluster design, configuration, and operations.
2. Sub-skills for **AKS troubleshooting** focused on common node health and networking issues.

Copilot uses AI. Check for mistakes.
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think you could take out line 27&28, since 30-36 covers the skills in details


The **AKS best practices** skill guides agents through cluster configuration recommendations across networking, upgrade strategy, security, reliability, scale, and more. The guidance reflects what the AKS engineering team recommends for production clusters for optimal performance and uptime, along with specific defaults and critical decisions that apply to AKS. After installing the skill, try the following commands to invoke the skill and receive targeted recommendations for your AKS cluster:

- "What are the best practice recommendations for a highly reliable and performant AKS cluster?"
- "Help me determine the Day-0 decisions I need to make when creating a new AKS cluster."
- "What networking setup is best for my AKS cluster?"

**AKS troubleshooting** covers the most common incident scenarios: node health failures and networking issues. It includes the exact CLI commands and diagnostic sequences that AKS engineers use internally when working customer incidents. Commands are permission-gated: the skill only suggests and executes commands that your current credentials allow, so there's no risk of unintentional changes. Here's what a typical troubleshooting session looks like in VS Code with the skill active:

```text
User: My node has been in NotReady for 10 minutes. What should I check?

Copilot [AKS troubleshooting skill]: Let's start with the node conditions and recent events.

> kubectl describe node <node-name>
> kubectl get events --field-selector involvedObject.name=<node-name> \
--sort-by='.lastTimestamp'

If the output shows a disk pressure or memory pressure condition, we'll check
the kubelet logs next. If it's a network issue, we'll look at the CNI plugin
and node connectivity. Share the output and we'll go from there.
```

## How to get started

You have two options: install individual skills directly, or get the full bundle through the GitHub Copilot for Azure plugin.

### Option 1: Install through GitHub Copilot for Azure plugin

The GitHub Copilot for Azure plugin is available through VS Code, Claude, and Copilot CLI and bundles AKS skills alongside 20+ skills across cost optimization, other Azure resources, and deployment workflows. Learn more about [GitHub Copilot for Azure](https://learn.microsoft.com/azure/developer/github-copilot-azure/introduction?tabs=vscode), and get the extension:

**Install the Azure plugin to VS Code:**
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you add some screenshots here?


1. Open VS Code and go to the Extensions marketplace (`Ctrl+Shift+X` / `Cmd+Shift+X`).
1. Search for **GitHub Copilot for Azure** and select **Install**. ([Direct link to the marketplace](https://marketplace.visualstudio.com/items?itemName=ms-azuretools.vscode-azure-github-copilot))
1. Open GitHub Copilot Chat in VS Code (the chat icon in the sidebar or `Ctrl+Alt+I` / `Cmd+Alt+I`).
1. Run a prompt that references AKS, such as *"What's the recommended upgrade strategy for my AKS cluster?"* or *"What are the best practices for AKS clusters?"*. The AKS skill will install and load automatically.

> **Note**: If skills don't activate after installing the extension, open the VS Code Command Palette (`Ctrl+Shift+P` / `Cmd+Shift+P`) and run **GitHub Copilot for Azure: Refresh Skills**. Alternatively, running any AKS-related prompt will trigger the extension to initialize and load available skills.

**Install the Azure plugin to Claude or Copilot CLI:**

1. Add the marketplace with `/plugin marketplace add microsoft/azure-skills`
1. Install the plugin with `/plugin install azure@azure-skills`
1. Update the plugin with `/plugin update azure@azure-skills`
1. Verify that the azure-kubernetes skill has successfully installed with `/skills`
1. Run a prompt that references AKS, which will invoke the relevant skill.

### Option 2: Install AKS skills directly

1. Go to [skills.sh](https://skills.sh/microsoft/github-copilot-for-azure) and locate the AKS skills: [azure-kubernetes](https://skills.sh/microsoft/github-copilot-for-azure/azure-kubernetes) and [azure-diagnostics](https://skills.sh/microsoft/github-copilot-for-azure/azure-diagnostics).
1. Follow the `npx skills add` command at the top of the page to install the skill directly.
1. Run any AKS-related prompt — for example, *"Review my cluster for best practices"* — and the skill will activate automatically.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

consider adding a section about how these aks skills differ or complement other aks ai capabilities? Like the AKS MCP server and agentic cli

## Conclusion

AKS skills give your agents a baseline of production AKS knowledge using the same guidance, commands, and diagnostic approaches that AKS engineers use. The first release covers best practices and troubleshooting, and we are planning to cover more scenarios based on customer feedback. If you run into issues or have scenarios you'd like to see covered, please feel free to open an issue on the [AKS repository](https://github.com/Azure/AKS/issues).
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you want to open up an issue specifically for aks skills and link that here and people can make requests on that issue directly so it's easier to track / upvote other people's requests?

5 changes: 5 additions & 0 deletions website/blog/tags.yml
Original file line number Diff line number Diff line change
Expand Up @@ -307,6 +307,11 @@ security:
permalink: /security
description: Securing clusters, workloads, supply chain, identities, and network boundaries in AKS.

skills:
label: Skills
permalink: /skills
description: Reusable agent skills, guided workflows, and domain-specific expertise for AKS tasks.

spot:
label: Spot
permalink: /spot
Expand Down
Loading