# kagent
*Also known as kagent*

- Website: https://kagent.dev
- Agent profile: https://directory.haycion.ai/agents/kagent-dev

> Kagent is an open-source CNCF sandbox project by Solo.io for building and running AI-driven agents on Kubernetes.

kagent is an open-source, CNCF sandbox project originated by Solo.io. It provides a cloud-native framework for building, deploying, and operating AI-driven agents within Kubernetes. The project includes a Kubernetes controller that manages agent lifecycles, an engine that runs the agent runtime, a dashboard and CLI for management, and a suite of MCP tooling to support Model Context Protocol servers and agent-enabled workflows. kagent supports running agents and MCP tools in Kubernetes, emphasizes safe, observable operations, and offers documentation to install, start, and extend its capabilities. The target audience includes developers building agent-based automation, site reliability engineers, and organizations seeking scalable, secure orchestration of autonomous capabilities in production. Its core concepts cover architecture components, runtimes, observability, and provider integrations.

**Mission:** To empower developers and operations teams to create, deploy, and manage autonomous agents within Kubernetes, enabling scalable, secure, and observable agent-based automation through MCP servers, tooling, and clear guidance.

## Products & Services

### [kagent](https://kagent.dev/)
*Platform*
Build and orchestrate AI agents in Kubernetes with efficient resource management and full observability.

- **Agent Substrate Runtime** — Ensures Fast Startups
- **Full Observability** — Ensures Transparency
- **Enterprise Distributions for kagent** — Delivers Context-Aware Distribution
- **Production-Ready Support** — Ensures Reliable Operations
- **Agent-to-Agent Workflows** — Facilitates Collaboration
- **CRD-Based Agent Lifecycle** — Streamlines Management
- **NVIDIA NemoClaw Integration** — Enhances Security
- **Advanced Management Features** — Streamlines Agent Management
- **Multi-Cluster Federation Support** — Facilitates Scalable Operations
- **Bring Your Own Frameworks** — Supports Diverse Frameworks
- **Go and Python ADK Runtimes** — Enhances Development Choices
- **Postgres Storage** — Provides Reliable Storage
- **Training and Technical Labs** — Facilitates Hands-On Learning
- **Context Compaction** — Reduces Token Usage
- **Long-Term Memory** — Improves User Experience
- **Prompt Templates** — Promotes Efficiency

### [kmcp](https://github.com/kagent-dev/kmcp)
*Product*
Accelerate MCP service development and deployment with kmcp's streamlined workflows.

- **Agentgateway Integration** — Secure Agent Communications
- **Kubernetes-native Deployment with kmcp Deploy** — Streamlined Deployment Process
- **Container Image Packaging with kmcp Build** — Simplify Image Management
- **MCP Services as CRDs** — Kubernetes Native Management
- **Project Scaffolding with kmcp Init** — Accelerated Project Setup
- **Multiple Transports Support** — Flexible Transport Options

## Market Segments

- **Autonomous Agent Orchestration** (market size $6.3B, CAGR 35.32%): Capabilities to compose, coordinate, and monitor multi-agent workflows, including memory and context management for autonomous AI agents.
- **Model Context Protocol deployment** (market size $900M, CAGR 35%): Tooling and deployment workflows for Model Context Protocol (MCP) services, including project scaffolding, container image packaging, CRD-based service management, support for stdio/HTTP transports, and agentgateway integration for secure tool communications.
- **AI governance and agent monitoring** (market size $430M, CAGR 34.3%): Monitors autonomous AI agents and models for policy compliance, performance, and risk; provides enforcement, audit capabilities, and analytics to govern AI-driven actions.
- **Federated and scalable runtime management** (market size $100M, CAGR 27.3%): Management of federated, multi-cluster Kubernetes deployments for AI runtimes, enabling multi-cluster federation, centralized control of agent fleets, production-grade Postgres-backed storage, and operational scalability.