Kagent is an open-source framework for building AI agents that run in Kubernetes. It brings cloud-native agentic AI to DevOps and platform engineering, so agents can help operate, troubleshoot, and automate the clusters they live in.
Kagent treats agents as Kubernetes-native resources: you declare agents, tools, and model configurations declaratively, and the framework runs them in the cluster with a CLI and a web UI for building and observing them. It integrates with the Model Context Protocol (MCP) so agents can use tools from the wider ecosystem, and it ships with prebuilt agents and tools aimed at common Kubernetes and cloud-native operations tasks. Kagent is a CNCF sandbox project, released under the Apache-2.0 license, with over 4,000 stars on GitHub and active development continuing through 2026. It works with the LLM provider of your choice, keeping you in control of models and data while agents assist with real operational work.
Kagent is free and open source under the Apache-2.0 license. You run it on your own clusters and pay only for your own infrastructure and model usage.
Platform engineers, SREs, and DevOps teams who want AI agents embedded in their Kubernetes operations for troubleshooting, automation, and day-to-day cluster tasks.