repository·main·Indexed 27 days ago
https://github.com/higress-group/higressA CNCF sandbox cloud-native gateway project built on Envoy and Istio, designed for high-performance traffic management and extensibility via Wasm plugins. Includes documentation for installation via Helm charts, configuration of gateway metrics, proxy settings, and Redis. Also features the hgctl CLI for creating, deploying, and managing AI Agents and Model Context Protocol (MCP) servers, with support for local deployment via AgentScope and cloud deployment via Alibaba Cloud Function Compute (AgentRun).
The MCP Server is an Envoy-based Golang Filter plugin for Higress that provides a unified Model Context Protocol (MCP) service interface. It enables integration with various backend services, including:
Requirement: You must use Higress version 2.1.0 or higher.
Higress is a cloud-native API gateway based on Istio and Envoy. It supports Wasm plugin development using Go, Rust, or JS and provides dozens of ready-to-use generic plugins and a management console.
Key Use Cases:
key-auth, hmac-auth, jwt-auth, basic-auth, and oidc.The GitHub MCP Server is an implementation of the Model Context Protocol (MCP) for the GitHub API. It enables AI agents to perform various GitHub operations including file manipulation, repository management, and advanced searching.
Key features include:
GET and POST, multi-turn dialogues, and both streaming and non-streaming modes. It also allows formatting results into a custom JSON structure using a provided schema.frontend-gray plugin implements frontend user canary/gray releases. It enables A/B testing and provides stability for system releases through strategies like canary deployment, monitoring, and rollback. It works by rewriting frontend resource requests (like HTML and JS files) to point to specific versions based on user identity or traffic weight.The cluster-key-rate-limit plugin provides cluster-level rate limiting using Redis. This ensures globally consistent rate limiting across multiple Higress Gateway instances.
It supports three modes:
rule_name and global_threshold.global_threshold (as a global fallback) and rule_items (per-dimension). In this mode, all matched rules are evaluated, and if any rule is triggered, the request is rejected.mcp-plate-quote tool integrates real-time market data for industry and concept sectors. It provides key indicators such as index levels, price change percentages, trading volumes, total market capitalization, sector rankings, and leading stock information. It is designed for intelligent investment research and tracking market trends through sector dynamics and constituent stock performance.The Higress RAG MCP Server provides tools for knowledge management and retrieval. The availability of tools depends on your configuration:
embedding and vectordb configuration.llm configuration.| Tool Name | Description | Dependencies | Required/Optional |
|---|---|---|---|
create-chunks-from-text | Splits text into chunks and stores them in a vector database for knowledge base construction | embedding, vectordb | Required |
list-chunks | Lists stored knowledge chunks for management | vectordb | Required |
delete-chunk | Deletes a specific knowledge chunk | vectordb | Required |
search | Performs semantic similarity search in the knowledge base | embedding, vectordb | Required |
chat | Answers user questions using RAG (Retrieval-Augmented Generation) | embedding, vectordb, llm | Optional |
The Brave Search MCP Server is an implementation that integrates the Brave Search API, providing the following capabilities:
frontend-gray plugin provides frontend user grayscale capabilities. It enables business A/B testing and ensures system release stability through grayscale, monitoring, and rollback strategies. It is particularly useful for micro-frontend architectures and CDN/OSS deployments.The Tool Search MCP Server is an implementation based on the Higress Golang Filter that provides semantic tool searching capabilities. It uses vector semantic search via a Milvus vector database and OpenAI-compatible Embedding APIs.
Key Features: