AI agents become much more useful when they can interact with real systems instead of relying only on what an LLM already knows. Model Context Protocol (MCP) provides a standardized way to connect agents with external tools, data, and services.

In this practical session, we’ll combine MCP, Google’s open-source Agent Development Kit (ADK), and Kubernetes to build an AI agent capable of investigating a cloud-native environment. Rather than simply asking an LLM, “Why is my application failing?”, we’ll allow the agent to use controlled tools to inspect Kubernetes resources, gather relevant context, and use Gemini to reason about what it discovers.

We’ll walk through the MCP client/server model, connecting MCP tools to an ADK agent, containerizing the components, and safely exposing Kubernetes capabilities to the agent. We’ll also discuss permissions and why agents should receive only the access they actually need.

**What You’ll Learn:**

* Understand MCP architecture
* Connect MCP tools to Google ADK
* Give agents controlled access to Kubernetes
* Inspect workloads and resources using an agent
* Containerize MCP and agent components
* Apply safe tool-access and permission patterns