Bringing Agentic AI to Kubernetes: Contributing Kagent to CNCF

Since announcing kagent, the first open source agentic AI framework for Kubernetes, on March 17, we have seen significant interest in the project. That’s why, at KubeCon + CloudNativeCon Europe 2025 in London, Solo.io is thrilled to announce its intent to contribute kagent to the Cloud Native Computing Foundation (CNCF) and bring AI agents to the cloud native ecosystem!

Kagent is a first-of-its-kind framework that helps DevOps and platform engineers build and run AI agents in Kubernetes and provides a foundation for AI-driven solutions in cloud native environments.

Within just two weeks of announcing the project, it has received:

  • Over 365 GitHub stars and counting
  • 135+ community members in the Discord channel
  • (and merged) 22 PR contributions

Why Agentic AI?

Agentic AI goes beyond the simple chat interactions provided by generative AI tools. It leverages advanced reasoning and iterative planning to tackle complex, multi-step problems autonomously, transforming insights into actions that drive productivity.

So, what happens when you apply agentic AI to cloud native environments? That’s where kagent comes in. It leverages this advanced reasoning to transform AI insights into concrete actions, helping platform and DevOps teams automate complex and time-consuming operational tasks like configuration, troubleshooting, observability, network security, and more. 

How does Kagent work?

As the first open source agentic AI framework for Kubernetes, kagent provides a catalog of agents, enabling anyone to run, build, and share AI-driven cloud native solutions. It is built on three key layers: tools, agents, and a declarative framework.

  • Tools: Any Model Context Protocol (MCP)-style function that agents can leverage to interact with cloud native systems. Kagent comes with pre-built tools with capabilities like displaying pod logs, querying Prometheus metrics, and generating resources.
  • Agents: Autonomous systems capable of planning and executing tasks, analyzing results, and continuously improving outcomes using one or more tools. Each agent can access one or more tools to accomplish its work or be grouped into teams where a planning agent assigns tasks to individual agents.
  • Framework: A simple declarative API and controller for building and running agents via UI, CLI, and declarative configuration. It is built on Microsoft's AutoGen framework and provides extensive customization options.

Kagent Roadmap

We’re also looking to the community to help us build out the kagent roadmap, creating an even more robust, seamless, and scalable solution. Planned areas of development include: 

  • Observability improvements: Agentic systems need to be able to introspect and understand the system's actions. In kagent, we plan to add even more powerful tracing capabilities, tighter integration with OpenTelemetry, and metrics for agents, LLMs, and tools. 
  • Feedback and testing: Agentic development is nondeterministic by nature. When an agent calls an LLM, it will receive an answer relevant to the context and prompt, while parameters and inputs can cause it to vary. For this reason, it's essential that kagent includes powerful feedback and testing strategies, such as debugging and time travel, evaluation frameworks, and guided learning.
  • Runtime and engine improvements: This includes expanding the core agent framework to accommodate a broader set of use cases. We plan to add capabilities, including multi-agent support, true declarative graph-style execution, and support for more workflows and LLM providers. 
  • Tools: Tools are one of the most important aspects of agentic systems. Because we are Kubernetes native, we are uniquely positioned to make tool use and discovery even easier. One way to expand on this is to expose our built-in tools as MCP servers to improve interoperability.

Why Cloud Native Computing Foundation?

Kagent was built for the cloud native community as an open source, Kubernetes-native way to harness the incredible potential of agentic AI. 

The project launched with tools for CNCF projects, including Argo, Helm, Istio, Kubernetes, and Prometheus, and we look forward to adding more to the tool registry with the help of the community. 

We believe that under vendor-neutral governance, the project’s adoption and ecosystem integration can flourish. Donating the project to the CNCF will expand the contributor base, foster innovation across organizations, and provide a comprehensive AI automation solution for cloud native end users across the globe. 

Get Involved

Kagent was designed to be extensible so that anyone can add their own agents and tools.

If you’re a platform or DevOps engineer, and leveraging AI agents to solve cloud native operation challenges excites you, or if you’re a developer or CNCF project maintainer interested in building AI agents to enrich our ecosystem, we’d love to collaborate.

Help us build and shape the future of AI-driven cloud native operations with kagent!

Cloud connectivity done right