@srbhptl39/mcp-superassistant-proxy
v0.1.8
Published
Run MCP SuperAssistant over SSE or visa versa
Maintainers
Readme
MCP SuperAssistant Proxy
MCP SuperAssistant Proxy lets you run multiple MCP stdio-based and SSE-based servers and expose them through a single SSE endpoint. This allows MCP SuperAssistant and other tools to connect to multiple remote MCP servers and tools via a unified proxy.
Installation & Usage
Run MCP SuperAssistant Proxy via npx:
npx -y @srbhptl39/mcp-superassistant-proxy@latest --config path/to/config.jsonCLI Options
--config, -c <path>: (required) Path to a JSON configuration file (see below)--port <number>: Port to run the proxy server on (default:3006)--baseUrl <url>: Base URL for SSE clients (default:http://localhost:<port>)--ssePath <path>: Path for SSE subscriptions (default:/sse)--messagePath <path>: Path for SSE messages (default:/message)--logLevel <info|none>: Set logging level (default:info)--outputTransport stdio | sse | ws | streamableHttp: Output MCP transport (default:ssewith--stdioor--config,stdiowith--sseor--streamableHttp)--ssePath "/sse": Path for SSE subscriptions (stdio→SSE, config→SSE modes, default:/sse)--messagePath "/message": Path for messages (stdio→SSE, stdio→WS, config→SSE, config→WS modes, default:/message)--streamableHttpPath "/mcp": Path for Streamable HTTP (stdio→StreamableHttp, config→StreamableHttp modes, default:/mcp)--stateful: Run StreamableHttp in stateful mode (stdio→StreamableHttp, config→StreamableHttp modes)--sessionTimeout 60000: Session timeout in milliseconds (stateful StreamableHttp modes only)--header "x-user-id: 123": Add one or more headers (stdio→SSE, SSE→stdio, or Streamable HTTP→stdio mode; can be used multiple times)--oauth2Bearer "some-access-token": Adds anAuthorizationheader with the provided Bearer token--logLevel debug | info | none: Controls logging level (default:info). Usedebugfor more verbose logs,noneto suppress all logs.--cors: Enable CORS (default:true)--healthEndpoint <path>: One or more endpoints returning"ok"(can be used multiple times)--timeout <ms>: Connection timeout in milliseconds (default:30000)
Configuration File
The configuration file is a JSON file specifying which MCP servers to connect to. Each server can be either a stdio-based server (run as a subprocess) or an SSE-based server (remote URL).
Example config.json
{
"mcpServers": {
"notion": {
"command": "npx",
"args": ["-y", "@suekou/mcp-notion-server"],
"env": {
"NOTION_API_TOKEN": "<your_notion_token_here>"
}
},
"gmail": {
"url": "https://mcp.composio.dev/gmail/xxxx"
},
"youtube-subtitle-downloader": {
"command": "bun",
"args": [
"run",
"/path/to/mcp-youtube/src/index.ts"
]
},
"desktop-commander": {
"command": "npx",
"args": ["-y", "@wonderwhy-er/desktop-commander"]
},
"iterm-mcp": {
"command": "npx",
"args": ["-y", "iterm-mcp"]
}
}
}- Each key under
mcpServersis a unique name for the server. - For stdio-based servers, specify
command,args, and optionallyenv. - For SSE-based servers, specify
url.
Endpoints
Once started, the proxy exposes:
- SSE endpoint:
http://localhost:<port>/sse - POST messages:
http://localhost:<port>/message - Streamable HTTP endpoint:
http://localhost:<port>/mcp - Websocket endpoint:
ws://localhost:<port>/message
(You can customize the paths with --ssePath and --messagePath.)
Example
- Create a config file (e.g.,
config.json) as shown above. - Run MCP SuperAssistant Proxy:
npx -y @srbhptl39/mcp-superassistant-proxy@latest --config config.json --port 3006
Why MCP?
Model Context Protocol standardizes how AI tools exchange data. If your MCP server only speaks stdio, MCP SuperAssistant Proxy exposes an SSE-based interface so remote clients (and tools like MCP Inspector or Claude Desktop) can connect without extra server changes. It also allows you to aggregate multiple MCP servers behind a single endpoint.
Advanced Configuration
MCP SuperAssistant Proxy is designed with modularity in mind:
- Supports both stdio and SSE MCP servers in one config.
- Automatically derives the JSON‑RPC version from incoming requests, ensuring future compatibility.
- Package information (name and version) is retransmitted where possible.
- Stdio-to-SSE mode uses standard logs and SSE-to-Stdio mode logs via stderr (as otherwise it would prevent stdio functionality).
- The SSE-to-SSE mode provides automatic reconnection with backoff if the remote server connection is lost.
- Health endpoints can be added for monitoring.
For more details, see the Model Context Protocol documentation.
Acknowledgments
Special thanks to SuperGateway for starter code.
