Kong Gateway is a cloud-native API gateway that also handles LLM and MCP traffic. It sits in front of your services as the central layer for proxying, routing, load balancing, health checking, and authentication, and it extends through plugins written in Lua, Go, or JavaScript.
For agent workloads, the AI Gateway capabilities present a universal LLM API that routes requests across providers such as OpenAI, Anthropic, Google Gemini, AWS Bedrock, Azure AI, Databricks, Mistral, and Hugging Face, and they add MCP traffic governance, security, and observability, including MCP server autogeneration from any RESTful API.
Kong runs in DB-backed, declarative database-less, or hybrid (separate control plane and data plane) deployment models, and runs natively on Kubernetes through its ingress controller. Configuration goes through a RESTful admin API, declarative files managed with decK, or the Kong Manager web UI. Konnect is the hosted control-plane offering.
Features
- Routing and load balancing: configurable routing, upstream load balancing, and health checks
- Authentication: JWT, basic auth, OAuth, ACLs, and other methods
- Universal LLM API: one interface across OpenAI, Anthropic, Gemini, Bedrock, Azure AI, Databricks, Mistral, Hugging Face, and more
- MCP governance: traffic control, security, and analytics for MCP servers, plus MCP generation from REST APIs
- AI plugins: 60+ AI features including observability, semantic security, semantic caching, and semantic routing
- Traffic plugins: rate limiting, request/response transformation, logging, and monitoring from the Plugin Hub
- Proxying: SSL/TLS termination and L4 or L7 connectivity
- Deployment models: declarative database-less and hybrid control-plane/data-plane setups
- Kubernetes: an ingress controller for serving Kubernetes clusters
