higress-group/higress

Health v2 · maintenance only

Adoption not part of the score

9196 stars · 1252 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Higress is a CNCF Sandbox, cloud-native AI-native API gateway built on Istio and Envoy, extensible with Wasm plugins written in Go, Rust, or JS. It unifies management of LLM API traffic, MCP Server hosting, and Kubernetes inference workloads with features like multi-model proxy, token rate limiting, semantic caching, and content safety.

Use cases

• proxy requests to multiple LLM providers behind one unified API
• host and expose MCP servers for AI agents
• rate-limit and track token usage per consumer
• cache LLM responses semantically to save tokens
• filter sensitive data and unsafe content in AI traffic
• serve as a Kubernetes ingress / API gateway
• fallback between model providers on failure

When to choose

• you need a production-grade gateway for LLM, agent, and MCP traffic
• you want protocol conversion across 100+ model providers with fallback
• you need token quotas, observability, and content safety at the gateway layer
• you want Envoy/Istio-based ingress with Wasm plugin extensibility

When to avoid

• you only need a simple reverse proxy without AI-specific features
• you cannot run Docker or Kubernetes infrastructure
• you need a lightweight client-side SDK rather than a gateway service

Facets

service · maturity active

api-gateway proxy rate-limiting caching middleware mcp llm-inference monitoring auth plugin-system large-language-models apis microservices cloud-computing self-hosted security cloud self-hosted go rust ai-gateway envoy istio wasm-plugins mcp-server-hosting token-management semantic-cache cncf-sandbox model-fallback ingress ai-agents devops docker kubernetes linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="higress-group/higress")