MCP HTTP Proxy
Exposes any stdio-based MCP server to the internet via HTTP/SSE transport, enabling remote agents to access MCP tools over a network.
README
MCP HTTP Proxy
Expose any stdio-based MCP server (like @browsermcp/mcp) to the internet via HTTP/SSE transport. Perfect for remote agents that need to access MCP tools over a network.
Features
- 🌐 Standard MCP SSE Transport - Follows MCP HTTP transport specification
- 🔐 API Key Authentication - Secure access with Bearer token
- 🚀 Multiple Exposure Options - Tailscale, ngrok, Cloudflare Tunnel
- 🔄 Session Management - Multiple concurrent sessions supported
- 📦 Works with any MCP server - Defaults to
@browsermcp/mcp
Quick Start
1. Install Dependencies
npm install
2. Configure (Optional)
cp .env.example .env
# Edit .env to set your API key
Generate a secure API key:
node -e "console.log(require('crypto').randomBytes(32).toString('hex'))"
3. Build and Run
npm run build
npm start
Or for development:
npm run dev
API Endpoints (Standard MCP SSE Transport)
| Endpoint | Method | Description |
|---|---|---|
/sse |
GET | Establish SSE connection (returns endpoint URL) |
/message?sessionId=xxx |
POST | Send JSON-RPC messages |
/health |
GET | Health check (no auth required) |
/sessions |
GET | List active sessions (auth required) |
Exposing to the Internet
Option 1: Tailscale (Recommended for Private Access)
Tailscale provides secure, private access without exposing to the public internet.
# Install Tailscale if not already installed
# Then serve via Tailscale
tailscale serve --bg --http 3000
Or use your Tailscale IP directly:
tailscale status # Find your Tailscale IP
Your MCP server will be available at:
https://your-machine.tailnet-name.ts.net(via serve)http://your-tailscale-ip:3000(direct IP)
Use from your agent:
{
"mcpServers": {
"browser-remote": {
"url": "http://your-tailscale-ip:3000/sse",
"headers": {
"Authorization": "Bearer your-api-key"
}
}
}
}
Option 2: ngrok (Public URL)
# Install ngrok if not already installed
ngrok http 3000
ngrok will provide a public URL like https://abc123.ngrok.io.
Option 3: Cloudflare Tunnel
# Install cloudflared if not already installed
cloudflared tunnel --url http://localhost:3000
How the SSE Transport Works
- Client GETs
/sse- Server creates session, sendsendpointevent with message URL - Client POSTs to
/message?sessionId=xxx- Server forwards JSON-RPC to MCP process - Server streams responses - MCP responses sent as SSE
messageevents
Usage Examples
Test Health Check
curl http://localhost:3000/health
Response:
{
"status": "healthy",
"sessions": 0,
"uptime": 10.5,
"transport": "sse",
"endpoints": {
"sse": "/sse (GET - establishes SSE connection)",
"message": "/message?sessionId=xxx (POST - send JSON-RPC messages)"
}
}
Connect via SSE (JavaScript)
const eventSource = new EventSource('http://localhost:3000/sse', {
headers: { 'Authorization': 'Bearer your-api-key' }
});
eventSource.addEventListener('endpoint', (e) => {
console.log('Message endpoint:', e.data);
// e.data = "/message?sessionId=xxx"
});
eventSource.addEventListener('message', (e) => {
console.log('MCP response:', JSON.parse(e.data));
});
Send MCP Request
curl -X POST "http://localhost:3000/message?sessionId=your-session-id" \
-H "Authorization: Bearer your-api-key" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
Configuration
| Environment Variable | Description | Default |
|---|---|---|
PORT |
Server port | 3000 |
HOST |
Host to bind to | 0.0.0.0 |
MCP_API_KEY |
API key for authentication | (none) |
MCP_COMMAND |
Command to run MCP server | npx |
MCP_ARGS |
JSON array of arguments | ["@browsermcp/mcp@latest"] |
Using Different MCP Servers
You can use this proxy with any stdio-based MCP server:
# Example: Use a custom MCP server
MCP_COMMAND=node
MCP_ARGS=["/path/to/your/mcp-server/dist/index.js"]
# Example: Use Python MCP server
MCP_COMMAND=python
MCP_ARGS=["-m", "your_mcp_server"]
Security Notes
- Always use API key authentication when exposing to the internet
- Use HTTPS (Tailscale, ngrok, and Cloudflare all provide this)
- Tailscale is recommended for private access within your tailnet
- Rotate API keys periodically
- Monitor sessions via
/sessionsendpoint
Troubleshooting
MCP Server Not Starting
Check the console logs for errors. The MCP server stderr is forwarded to the proxy's console.
Connection Refused
- Ensure server is running:
curl http://localhost:3000/health - Check Tailscale connectivity:
tailscale status
Authentication Failed
- Check the API key in
.envmatches your request header - Ensure "Bearer " prefix is included in Authorization header
SSE Not Working
- Ensure you're connecting to
/sseendpoint (not just base URL) - Check that Authorization header is sent with GET request too
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.