Higress is a cloud-native API gateway based on Istio and Envoy that emphasizes AI-native capabilities and extensibility through WebAssembly plugins.
The tool addresses the need for unified API management across traditional services and AI workloads. It originated at Alibaba to solve specific problems including long-connection disruption during gateway reloads and improved load balancing for gRPC and Dubbo protocols. Higress extends the Envoy proxy with WebAssembly plugins written in Go, Rust, or JavaScript, providing dozens of ready-to-use general-purpose plugins alongside AI-specific functionality. Its AI gateway capabilities support mainstream model providers both domestic and international, and it can host Model Context Protocol servers, enabling AI agents to call various tools and services. The project includes an openapi-to-mcp tool for converting OpenAPI specifications into remote MCP servers.
Higress suits organizations running cloud-native infrastructure who need to manage both traditional API traffic and AI model access through a single gateway. It is particularly relevant for teams building AI agent systems that require integration with multiple external services and LLM providers. The tool offers an out-of-the-box console for management and can be started with Docker for local development. Teams already invested in Istio and Envoy ecosystems will find natural integration points, though Higress functions as a standalone gateway as well.
The project maintains active development with regular updates to its AI provider integrations and plugin ecosystem. The codebase shows ongoing refinement of core gateway functionality alongside expansion of AI-specific features. Documentation is comprehensive, covering architecture, plugin development, and deployment patterns. The project operates as a vendor-neutral CNCF initiative with adoption across multiple industries documented in its community materials.
