Connect other MCP clients

Before you start

1. Open Gadwal and click AI & MCP in the sidebar of the Connections window (⌘2). The same page is also a tab in Gadwal › Settings (⌘,).
2. Turn on MCP server. The Address line shows http://127.0.0.1:55183/mcp (you can change the port below it). The server only listens on this Mac.
3. Check each connection's AI level: Off, Read-only, Ask before write or Full. Production connections start at Read-only. The AI only sees connections that aren't Off. See AI levels and hidden columns.
4. Keep Gadwal open. The server runs inside the app, so AI tools stop working when you quit it.

AI tools need an active trial or the lifetime unlock. After the trial, turning the server on opens Unlock, and a server that is still running answers every tool call with a message saying so.

In Gadwal

Scroll to the Other row under AI Apps on the AI & MCP page (below Cursor). Its button reads Copy JSON, and How to connect under it lists the same steps as this page.

Settings to use

• Transport: Streamable HTTP
• URL: http://127.0.0.1:55183/mcp
• Header: Authorization: Bearer <token>

The JSON block

On the AI & MCP page, under AI Apps, scroll down and click Copy JSON on the Other row (below Cursor). It copies this block with your token, in the mcpServers format many clients use:

{
"mcpServers": {
  "gadwal": {
    "type": "http",
    "url": "http://127.0.0.1:55183/mcp",
    "headers": { "Authorization": "Bearer <token>" }
  }
}
}

Add it to your app’s MCP settings as a Streamable HTTP server. If the app can’t read this format, use the three settings above.

Clients that only start stdio servers need a bridge to reach an HTTP server. Clients running in the cloud can’t reach your Mac at all.