Tiangong Ai MCP

🚀 Tiangong AI Model Context Protocol (MCP) Server

The Tiangong AI Model Context Protocol (MCP) server supports three protocols: STDIO, SSE (deprecated), and Streamable Http.

🚀 Quick Start

📦 Installation

Client STDIO Server

npm install -g @tiangong-ai/mcp-server

npx dotenv -e .env -- \
npx @tiangong-ai/mcp-server

Remote SSE Server

npm install -g @tiangong-ai/mcp-server
npm install -g supergateway

npx dotenv -e .env -- \
npx -y supergateway \
    --stdio "npx -y -p @tiangong-ai/mcp-server tiangong-ai-mcp-stdio" \
    --port 3001 \
    --ssePath /sse --messagePath /message

Using Docker

# Build the MCP server image using Dockerfile (optional)
docker build -t linancn/tiangong-ai-mcp-server:0.0.8 .

# Pull the MCP server image
docker pull linancn/tiangong-ai-mcp-server:0.0.8

# Start the MCP server using Docker
docker run -d \
    --name tiangong-ai-mcp-server \
    --publish 3001:80 \
    --env-file .env \
    linancn/tiangong-ai-mcp-server:0.0.8

💻 Development

Environment Setup

# Install Node.js
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.2/install.sh | bash
nvm install 22
nvm use

# Install dependencies
npm install

# Update dependencies
npm update && npm ci

Code Formatting

# Format the code using the code checker
npm run lint

Local Testing

STDIO Server

# Start the STDIO server using MCP Inspector
npm run start

SSE Server

# Package the current project
npm run build && npm pack

# Globally install supergateway if needed (optional)
npm install -g supergateway

# Start the SSE server. If the --baseUrl parameter is configured, it should be set to a valid IP address or domain name
npx dotenv -e .env -- \
npx -y supergateway \
    --stdio "npx -y -p tiangong-ai-mcp-server-0.0.8.tgz tiangong-ai-mcp-stdio" \
    --port 3001 \
    --ssePath /sse \
    --messagePath /message

# Start MCP Inspector
npx @modelcontextprotocol/inspector

Publishing

npm login

npm run build && npm publish