Skip to content

AKS MCP with GHCP Auto-detect #808

Description

@willspacecase

Feature Summary

Automatically configure and enable AKS MCP when GitHub Copilot is detected and enabled as the AI provider, allowing users to access the same AI troubleshooting experience available in AKS desktop without additional MCP setup.

Motivation

GitHub Copilot can be automatically detected and configured as an AI provider, but AKS MCP currently requires separate manual setup.

This creates additional setup steps before users can access the same AI troubleshooting experience available in AKS desktop. Users must understand MCP concepts and manually configure an MCP server before they can use AKS-specific troubleshooting capabilities.

The goal is to make AKS MCP a natural extension of the GitHub Copilot auto-detect experience so users can immediately begin troubleshooting AKS clusters without additional configuration.

Detailed Description

When GitHub Copilot is detected and enabled, the AI Assistant should automatically configure and enable AKS MCP as part of the default experience.

GitHub Copilot provides the model experience, while AKS MCP provides access to AKS-specific tools and cluster context. From a user's perspective, these should work together as a single integrated experience.

The default experience should:

  • Detect when GitHub Copilot is available and enabled.
  • Automatically configure AKS MCP.
  • Use a safe default access level of read-only.
  • Reuse the user's existing Azure authentication and RBAC permissions.
  • Eliminate the need for manual MCP server setup for the common AKS desktop workflow.
  • Continue to support manual MCP configuration for advanced scenarios.

Alternatives Considered

Continue with manual MCP configuration

Users manually add and configure the AKS MCP server.

Pros

  • No additional implementation work.
  • Maintains full user control.

Cons

  • Creates friction during onboarding.
  • Requires users to understand MCP concepts and configuration.
  • Results in a different experience than the AI troubleshooting workflow available through AKS agentic cli.

Use Cases

Describe the use cases where this feature would be helpful:

  1. As an AKS desktop user, I want AKS MCP automatically configured when I enable GitHub Copilot so that I can immediately troubleshoot AKS clusters without additional setup.
  2. When GitHub Copilot is detected and enabled, I need AKS MCP to be automatically available so I can access AKS-specific troubleshooting capabilities from the AI Assistant.
  3. As a new user, I want the AI Assistant to work out of the box so that I do not need to learn MCP concepts or configure servers manually.

Implementation Ideas

  • Extend the GitHub Copilot auto-detect flow to also configure AKS MCP.
  • Create a default AKS MCP configuration using:
  • Command: aks-mcp
  • Transport: stdio
  • Access level: readonly
  • Reuse existing Azure authentication and RBAC permissions.
  • Validate prerequisites during initialization and provide actionable guidance if requirements are missing.

Additional Context

Given out target audience of new k8's users, natural language troubleshooting is a core AKS desktop experience. Users should not need to understand MCP concepts or perform additional configuration steps to access AKS-specific AI troubleshooting capabilities.

The combination of GitHub Copilot and AKS MCP should feel like a single integrated experience from the user's perspective.

Acceptance Criteria

  • GitHub Copilot detection automatically enables and configures AKS MCP.
  • Users are not required to manually add an MCP server, copy JSON configuration, or enter command-line arguments.
  • AKS MCP uses a read-only access level by default.
  • AKS MCP respects the user's existing Azure authentication and RBAC permissions.
  • Clear and actionable guidance is provided when required prerequisites are missing.
  • Manual MCP configuration remains available for advanced scenarios.

Checklist

  • I have searched for existing feature requests
  • I have provided clear use cases
  • I have considered alternatives
  • This feature aligns with the project's goals

Metadata

Metadata

Assignees

Labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions