Connect ChatGPT, Claude, or Gemini tooling

Before you connect

1. Sign in to the owner control centre and complete Umbraco MFA.
2. Install and pair at least one Windows or macOS computer.
3. Keep that computer awake and approve remote work in its local Umbraco controls.
4. Use only the official endpoint below. Do not share pairing files, owner passwords, recovery codes, or controller tokens with an AI client.

https://umbraco.blueskyadvisory.ae/mcp

Recommended

ChatGPT and Codex plugin

1. Open Plugins in ChatGPT on the web or in the ChatGPT desktop app.
2. Find Umbraco.AI in the workspace, personal, or shared marketplace and select the plus button to install it. If it is not listed, ask the deployment owner to share or publish its marketplace entry.
3. Select Connect when prompted. The plugin connects to Umbraco's hosted MCP service.
4. In the browser window, sign in to Umbraco, complete MFA, review the requested permissions, and approve them.
5. Start a new chat, type @Umbraco.AI, and ask: “List my paired computers.”

Installation and connection are separate: installing the plugin does not grant it access until you finish Umbraco sign-in and approve its scopes. Managed workspaces may require an administrator to make the plugin available.

Official ChatGPT plugin instructions

Claude custom connector

Pro or Max

1. In Claude, open Customize → Connectors.
2. Select + → Add custom connector, name it Umbraco, and paste the hosted MCP address above.
3. Leave the optional OAuth client ID and client secret empty. Umbraco registers public clients dynamically.
4. Add the connector, then complete Umbraco sign-in, MFA, and permission approval.
5. In a conversation, open + → Connectors, enable Umbraco, and ask Claude to list your paired computers.

Team or Enterprise

An Owner or Primary Owner must first add the URL under Organization settings → Connectors → Add → Custom → Web. Each member then connects and authorizes their own access.

Official Claude custom-connector instructions

Advanced developer setup

Gemini CLI

The Gemini consumer web and mobile app do not have a supported Umbraco installation path. For Gemini CLI, use Umbraco's local stdio gateway. Do not paste the hosted URL into Gemini CLI yet: its browser OAuth callback is not part of the current production acceptance matrix.

1. Clone the Umbraco repository and install the optional MCP dependencies in a dedicated virtual environment.
2. Copy examples/gateway.example.json. Set the HTTPS relay URL and an absolute path to a controller-token file; never put the token itself in client settings.
3. Copy examples/clients/gemini.settings.json into the appropriate user or project settings. Replace the Python executable and gateway-config path with absolute paths.
4. Keep trust set to false, then run gemini mcp list and use /mcp to review the discovered tools.
5. Test with “List my paired computers” before enabling any write, terminal, file, or application-launch permission.

Official Gemini CLI MCP instructions · Umbraco gateway reference

Verify and remove access

1. Ask the assistant to list computers. A successful result should show only devices paired to this deployment.
2. Run a read-only system-information check before approving broader capabilities.
3. Review assistant permissions and activity in the owner control centre.
4. When access is no longer needed, revoke the assistant in Umbraco and disconnect or uninstall it in the AI client.

A connected assistant cannot wake or control an offline computer unless the computer, router, recorded public IP, Wake-on-LAN settings, and local approval controls are all configured for that action.