Moonshot MCP Server Gateway

Moonshot MCP Server Gateway provides a lightweight, unified entry point to access multiple MCP servers. It simplifies connection and management by routing client connections to the appropriate MCP sub-servers and supporting multiple transport protocols for flexible deployments.

How to use

Connect your MCP client to the gateway using the gateway’s host and port, then specify the sub-server you want to reach by using the API routing prefix configured for that sub-server. The gateway handles protocol translation and load-balancing across the configured MCP sub-servers, so you can access different MCP services through a single, centralized entry point.

Practical patterns you can use: use the HTTP-based remote MCP URL as the primary entry point when you want centralized routing to remote MCP services. If you prefer local execution, configure a local MCP sub-server and connect via the gateway using the designated API prefix. In all cases, ensure your client targets the gateway address and uses the sub-server prefix to route requests to the intended MCP service.

How to install

Prerequisites you need before installation: a machine with Python 3.8+ and Git installed. You will also need a shell with standard commands like git, cp, and a Python-compatible environment tool.

Step 1: Clone the gateway repository and prepare the environment.

git clone https://github.com/MarshallEriksen-shaomingyang/moonshot-mcp-server.git
cp .env.example .env
cp moonshot_config.example.toml moonshot_config.toml
```} ,{

Step 2: Create and activate a Python virtual environment, then install any required dependencies via your project’s package manager. The recommended run flow uses the bundled tooling to start the gateway in a chosen mode.

# Create virtual environment
uv venv
# Activate virtual environment
source .venv/bin/activate

Step 3: Start the gateway server in the mode you need. You can run in HTTP mode or SSE mode as shown.

uv run server.py --mode http
# or for SSE mode
uv run server.py --mode sse

Configuration and notes

The gateway configuration allows you to define a basic server plus multiple MCP sub-servers. Each sub-server has a unique name and a routing prefix used to reach that sub-server through the gateway.

Example configuration structure you can adapt for your environment:

# Basic server configuration
[server]
name = "AlphaCore Server Gateway"
version = "1.0.0"
port = 8090
host = "0.0.0.0"

# HTTP-based MCP sub-server example
[mcpServers.http_server]
type = "http"
url = "https://api.example.com/mcp"
prefix = "api"

"mcpServers" can also include local/process-based sub-servers when your setup requires them, but a minimal, explicit HTTP sub-server example is shown above. Ensure every HTTP sub-server includes a valid url and a prefix, and every local/sub-server entry includes its required fields according to its type.

## Available tools

### Awesome MCP Server List

A curated list of available MCP servers and examples to help you discover integrations.

### mcp.so

An online resource for MCP server discovery, tooling, and examples.

### glama

A tooling resource for MCP servers that provides insights and quick-start guidance.