Kong is a cloud-native gateway that proxies and governs conventional API traffic, microservices, and agentic LLM and MCP traffic. It provides a platform-agnostic proxy layer with plugin extensibility, declarative deployment modes, and traffic governance for AI and MCP integrations.
Project overview
The gateway applies centralized governance, security, observability, and routing to both agentic MCP traffic and conventional APIs, with documented support for declarative and hybrid control-plane/data-plane deployment models.
Project type
Infrastructure
Deployment
Refer to project documentation
License
Apache-2.0
Best for
Developers and operations teams who need a platform-agnostic proxy layer to centralize functionality and governance across microservices, conventional APIs, and agentic LLM and MCP traffic.
Teams seeking deployment flexibility for their API infrastructure using declarative databaseless or hybrid control-plane and data-plane models.
Key capabilities
Centralizes common API, AI, and MCP functionality across services to act as a proxy layer.
Provides MCP traffic governance, security, observability, and autogeneration from any RESTful API.
Provides 60+ AI features including AI observability, semantic security and caching, and semantic routing.
Supports declarative Databaseless Deployment and Hybrid Deployment (control plane/data plane separation) without vendor lock-in.
Limitations and risks
GPU requirements, minimum hardware, operating system constraints, and database dependencies are not documented in the provided configuration.
Specific telemetry collection practices and configurations are not documented in the provided facts.
Getting started
Setup difficulty is undocumented. Administrators can manage and interact with the system using a web GUI or an API. No coding is strictly required for basic operation.
Alternatives and comparisons
Provides a unified interface to call 100+ LLM providers with cost tracking, guardrails, load balancing, and an admin dashboard.
An OpenAI-compatible AI gateway offering automatic failover, load balancing, and semantic caching across multiple providers.
Aggregates and accesses various LLM and media generation providers through a unified interface.