AI agents that live inside your cluster — defined as Kubernetes resources, armed with real tools, and ready to investigate and operate your infrastructure.

Welcome to kagent: Cloud-Native Agentic AI for Kubernetes.

kagent is a CNCF sandbox project, originally created by Solo.io, that brings AI agents into Kubernetes as first-class citizens. Instead of running an agent as a script on someone's laptop or a bespoke service wired together by hand, kagent lets you declare an agent the same way you declare a Deployment: a YAML manifest, version-controlled in Git, reviewed in a pull request, and reconciled by a controller. Each agent is backed by a large language model, connected to real tools through the Model Context Protocol, and capable of investigating incidents, querying metrics, and operating the cluster on your behalf. This book walks the whole system from the ground up. The opening chapters explain why DevOps and SRE teams need agents that live where the infrastructure lives, and lay out kagent's architecture — the controller, the engine, the API server, the UI, and the CLI. From there we dig into the custom resources that define an agent: the Agent itself, the ModelConfig that points at an LLM, and the ToolServer and RemoteMCPServer resources that wire in tools. Later chapters cover how agents actually run, how they talk to each other through the A2A protocol, how to keep humans in the loop, how to observe and secure them, and finally how to put it all into practice on a real cluster.

This title is part of the ShriIra library and is free to read in full, right here — our small contribution to making world-class knowledge easy to reach.

A note on reading it: open the Contents menu at the top of the reader to jump between chapters, use the Aa menu to set a comfortable text size, theme (light, sepia, or night), and single- or two-page layout. Your place is saved automatically, so you can always pick up where you left off.

We hope it serves you well.

— Shriira Press

Contents

1. Chapter 1 — Agents That Live in the Cluster
2. Chapter 2 — Architecture
3. Chapter 3 — The Agent Resource
4. Chapter 4 — Models and ModelConfig
5. Chapter 5 — Tools and the Model Context Protocol
6. Chapter 6 — Multi-Agent Workflows and A2A
7. Chapter 7 — Observability, Security, and Human-in-the-Loop
8. Chapter 8 — kagent in Practice