Chosen by teams shaping the future of AI and cloud infrastructure
Closing the gaps between AI agents and production
Kubernetes as a first-class agentic runtime
kagent natively extends Kubernetes to support agents, tools, and skills as first-class entities, with the identity, policy, sandboxing, and observability controls production demands.
One gateway for every agentic protocol
Prior-generation proxies weren't built for agentic protocols. agentgateway was, from the ground up, with full MCP and A2A implementation. Model aliasing, token-weighted rate limiting, intelligent routing. One gateway, not three layered on top of each other.
Control that doesn't slow teams down
Developers pick the framework, the model, the tool. Control moves to the platform layer. MCP server provenance checks, approved-catalog workflows, end-to-end tracing, centralized policy. Teams move fast. Nothing runs unchecked.
Our commitment to the open source community
Our support for the open-source landscape runs deep, from our donation of projects like kagent and agentgateway to our recognition as a top 10 CNCF contributor.
GitHub Stars
Community Members
Downloads
CNCF Contributor
agentregistry
One registry for MCP servers, agents, skills, and prompts.
kagent
Bringing Agentic Al to Cloud Native.
agentgateway
One high-performance gateway for service, LLM, and MCP traffic.
agentdesktop
Visibility, security, and control for agents where they actually run - on the desktop.