Using an OpenMCP server

Every server in the registry is available as a remote endpoint and is also published on github and npm to download and run locally

Using a remote server

To use a hosted Streamable HTTP server, add the following to your client config:

{
  "mcpServers": {
    "{server-id}": {
      "transport": "streamableHttp",
      "url": "https://mcp.open-mcp.org/api/server/{server-id}@latest/mcp"
    }
  }
}

Replace {server-id} with an appropriate value e.g. autobahn

You can forward "environment" variables to the remote server by including them in the request headers or URL query string (headers take precedence). Just prefix the variable name with FORWARD_VAR_ like so:

https://mcp.open-mcp.org/api/server/{server-id}@latest/mcp?FORWARD_VAR_OPEN_MCP_BASE_URL=https%3A%2F%2Fapi.example.com

Security

Sending authentication tokens as forwarded variables is not recommended

Installing locally

If you want to run a server locally on your own machine instead of using the hosted version, use the CLI installer:

Claude desktop

npx @open-mcp/config add {server-id} \
  ~/Library/Application\ Support/Claude/claude_desktop_config.json \
  --ENV_VAR=abc123

Cursor

Run this from the root of your project directory or, to add to all cursor projects, run it from your home directory ~

npx @open-mcp/config add {server-id} \
  .cursor/mcp.json \
  --ENV_VAR=abc123

Other

npx @open-mcp/config add {server-id} \
  /path/to/config.json \
  --ENV_VAR=abc123

Manually

If you don't want to use the command-line helper, add the following to your MCP client config manually:

{
  "mcpServers": {
    "{server-id}": {
      "command": "npx",
      "args": ["-y", "@open-mcp/{server-id}"],
      "env": {...}
    }
  }
}