Overview
kagent is a Kubernetes-native framework that brings agent, tool and model configuration into Kubernetes via CRDs. It includes a controller, engine, UI and CLI to simplify deploying, observing and debugging AI agents in clusters.
Key Features
- Kubernetes-native CRDs (Agent, ToolServer, etc.)
- Multi-LLM provider support (OpenAI, Azure, Anthropic, Vertex, Ollama, …)
- Built-in MCP tools, memory and OpenTelemetry observability
Use Cases
- Deploy managed conversational/agent services in Kubernetes
- Expose external tools to agents via MCP
- Run agent-driven automation in CI/CD workflows
Technical Highlights
- Declarative resource management with hot updates and observability
- Provides UI and CLI for local dev and cluster workflows
- Extensible plugin and tool ecosystem