Kagent

Kagent: Open-Source AI Agent Framework for Kubernetes

Kagent is a CNCF Sandbox open-source framework donated by Solo.io that enables teams to build and run AI agents directly within Kubernetes environments. Built on the AutoGen multi-agent framework and supporting the Model Context Protocol (MCP), Kagent provides the infrastructure for custom AI-powered automation in cloud-native stacks.

Building AI Agents for Kubernetes

Kagent provides a Kubernetes Operator and a set of pre-built tools for common cloud-native operations. Teams can define AI agents that have access to Kubernetes API, Helm, Argo CD, Istio, Grafana, Prometheus, and more — enabling automation workflows that were previously only possible with custom scripting or expensive commercial tools.

Key Features

• Kubernetes-native deployment: Agents run as Kubernetes CRDs and operators
• MCP support: Model Context Protocol for connecting to AI assistant ecosystems
• AutoGen framework: Multi-agent orchestration for complex workflows
• Pre-built tool catalog: Kubernetes, Helm, Argo CD, Istio, Grafana, Prometheus integrations
• Custom agents: Build domain-specific agents for your stack
• CNCF governance: Vendor-neutral project under CNCF Sandbox
• Open source: Full Apache 2.0 license

Who Uses Kagent?

Kagent is for platform engineers and DevOps practitioners who want to build custom AI-powered automation for their Kubernetes environments without vendor lock-in. It's a building block rather than a finished product — ideal for teams with the engineering capacity to build custom AI workflows.

Pricing

Kagent is free and open source under Apache 2.0.