> For the complete documentation index, see [llms.txt](https://help.intangible.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.intangible.ai/intangible-mcp-agents/google-gemini.md).

# Google Gemini

Intangible works with Google Gemini: the Gemini app on personal Google accounts, and Gemini CLI. Apps, models and setup steps.

**Yes, Intangible works with Google Gemini.** The Gemini app and Gemini CLI connect to Intangible's MCP server. Both build 3D scenes in your Studio tab from a plain-language description.

## At a glance

| Google app                            | Where you add Intangible                                      | Sign-in                                                      |
| ------------------------------------- | ------------------------------------------------------------- | ------------------------------------------------------------ |
| Gemini app (personal Google accounts) | **Customize › Connectors › + Custom**                         | **Connect**, **Agree and continue**, then **Approve access** |
| Gemini CLI                            | `~/.gemini/settings.json`, under `mcpServers`, with `httpUrl` | `/mcp auth intangible`                                       |

## Apps and models

| Google app                                               | Model                                 |
| -------------------------------------------------------- | ------------------------------------- |
| Gemini app (gemini.google.com), personal Google accounts | Gemini's default model                |
| Gemini CLI                                               | Gemini 4 Argon (`gemini-4-argon`)     |
| Gemini CLI                                               | Gemini 3.8 Flash (`gemini-3.8-flash`) |

{% hint style="warning" %}
**Work and school Google accounts can't add Intangible to the Gemini app.** This is Google's limit. The Gemini app adds an MCP server as a custom connected app. Google allows custom connected apps only on personal Google accounts. On a Google Workspace account, Connected Apps has no **Custom** button, and your Workspace admin can't turn it on. Sign in to the Gemini app with a personal Google account, or use Gemini CLI. Gemini CLI works with work accounts. Google also requires you to be 18 or over and in the US, with **Keep Activity** on. See Google's [Connect & manage custom apps for Gemini Apps](https://support.google.com/gemini/answer/17209137).
{% endhint %}

## Before you start

* An Intangible account. Sign up at [studio.intangible.ai](https://studio.intangible.ai/?utm_source=help.intangible.ai\&utm_medium=docs\&utm_campaign=inline-link\&utm_content=google-gemini).
* For the Gemini app: a personal Google account, with Keep Activity on.
* For Gemini CLI: Node.js with npm, and a terminal.

## Set it up

The server URL is `https://mcp.intangible.ai/mcp`.

{% tabs %}
{% tab title="Gemini app" %}

### Gemini app: add Intangible as a custom app

1. Go to [gemini.google.com](https://gemini.google.com) signed in with a **personal** Google account. Check the account in the bottom left: a work account shows **Work** under your name.
2. Make sure **Keep Activity** is on at [myactivity.google.com/product/gemini](https://myactivity.google.com/product/gemini). Google hides custom apps when it's off.
3. In the sidebar, click **Customize**, then the **Connectors** tab.
4. Click **+ Custom** at the top right.
5. Under **MCP Server URL**, enter `https://mcp.intangible.ai/mcp`. Leave **Additional settings** (Client ID and Client secret) empty. Click **Next**.
6. A dialog reads **Connect this MCP server**. Scroll to the bottom, tick **I understand and accept the security and privacy risks of connecting this custom app**, then click **Connect**.
7. A second dialog names **\[Custom] Intangible**. Scroll down and click **Agree and continue**.
8. An Intangible page opens. Click **Approve access**.

The Google account and the Intangible account don't have to match. Gemini needs a personal Google account. The Intangible sign-in can be any Intangible account, including a work one. Gemini builds in the Intangible account you approve.

### Gemini app: build a scene

1. Open the project you want to build in at [studio.intangible.ai](https://studio.intangible.ai/?utm_source=help.intangible.ai\&utm_medium=docs\&utm_campaign=inline-link\&utm_content=google-gemini). Close every other Studio tab first, including other projects and the Studio home page.
2. In Gemini, click **New chat**.
3. Type `@` and pick **Python Mcp Server** from the list. That's Intangible (see the naming note below).
4. After the chip, name the project and ask Gemini to confirm it before building, then [describe your first scene](/intangible-mcp-agents/intangible-mcp-agents.md#describe-your-first-scene). For example:

   > My Studio tab is open on my Intangible project Lobby Concept. First call get\_or\_set\_studio\_target and confirm the project name. Then build a small living room: a sofa facing a TV on a low stand and a coffee table between them. Don't delete anything. Use models from Intangible's asset library.
5. Gemini asks before every tool, for example *Let Gemini use "run scene script" from Python Mcp Server*. Read the request, then click **Allow**.

{% hint style="warning" %}
**One Studio tab, and say which project.** With more than one Studio tab open, Gemini can build in the wrong one. A script can also clear that scene before it builds. Close the other tabs, name the project in your message, and tell Gemini not to delete anything. Before you click **Allow** on a **run scene script** request, check that the script doesn't start by deleting objects.
{% endhint %}

Gemini may block out the scene with simple shapes. Ask it to use models from Intangible's asset library if you want real objects.

{% hint style="info" %}
**Intangible goes by other names inside Gemini.** Google labels the server from the server's own data. You'll see:

* **\[Custom] Intangible** on the consent dialog when you connect.
* **Python Mcp Server** in the chat's `@` list and in each tool request. Pick this one.
* **\[Custom] Python Mcp Server** under your Google Account's [Linked apps](https://myaccount.google.com/linkedapps).
* Nothing under **Active** on the Connectors page, even when it's connected. Use the `@` list instead.

Don't add Intangible again because it's missing from **Active**. It's already connected.
{% endhint %}
{% endtab %}

{% tab title="Gemini CLI" %}

### Gemini CLI: install

1. In a terminal, install Gemini CLI:

   ```bash
   npm install -g @google/gemini-cli
   ```
2. Confirm it installed:

   ```bash
   gemini --version
   ```

### Gemini CLI: add the Intangible server

1. Open `~/.gemini/settings.json` in a text editor. Create the file if it doesn't exist.
2. Add an `intangible` entry under `mcpServers`. If the file already has settings, keep them and add `intangible` alongside any other servers:

   ```json
   {
     "mcpServers": {
       "intangible": {
         "httpUrl": "https://mcp.intangible.ai/mcp",
         "oauth": { "enabled": true },
         "timeout": 120000
       }
     }
   }
   ```

   Use `httpUrl`. Gemini CLI reads `url` as an older server type, and the connection doesn't complete.
3. Save the file.

### Gemini CLI: sign in

1. Start Gemini CLI:

   ```bash
   gemini -m gemini-4-argon
   ```
2. The first time, Gemini CLI asks how to sign in to Google: **Sign in with Google**, **Use Gemini API key** or **Vertex AI**. Pick the one that matches your Google account or plan.
3. Sign in to Intangible:

   ```
   /mcp auth intangible
   ```
4. A browser tab opens on an Intangible page that names Gemini CLI. Click **Approve access**.
5. Run `/mcp`. **intangible** shows as connected, with its tools listed.

### Gemini CLI: build a scene

1. Open a project in Studio at [studio.intangible.ai](https://studio.intangible.ai/?utm_source=help.intangible.ai\&utm_medium=docs\&utm_campaign=inline-link\&utm_content=google-gemini). Keep only that one Studio tab open.
2. In Gemini CLI, ask it to use Intangible, then [describe your first scene](/intangible-mcp-agents/intangible-mcp-agents.md#describe-your-first-scene).
3. When Gemini CLI asks to run an Intangible tool, choose to allow it always, or it asks before every step.

To use Gemini 3.8 Flash instead, start with `gemini -m gemini-3.8-flash`, or switch inside a session with `/model`.
{% endtab %}
{% endtabs %}

## Setup notes

* **Vertex AI sign-in.** To sign in through a Google Cloud project, put these lines in `~/.gemini/.env`, with your own project ID, and run `gcloud auth application-default login` first:

  ```bash
  GOOGLE_GENAI_USE_VERTEXAI=true
  GOOGLE_CLOUD_PROJECT=your-project-id
  GOOGLE_CLOUD_LOCATION=global
  ```
* **"invalid\_grant" or "invalid\_rapt".** Your Google Cloud sign-in expired. Your Google Workspace sets how long it lasts. Run `gcloud auth application-default login` again, then restart Gemini CLI. Your Intangible sign-in is kept.
* **Teams.** If you belong to more than one Intangible team, Gemini asks which team to create a project in.

## Primary source

Google documents Gemini CLI's MCP setup here: [MCP servers with Gemini CLI](https://geminicli.com/docs/tools/mcp-server/).

## Related

* [Agents & MCP](/intangible-mcp-agents/intangible-mcp-agents.md)
* [Choosing a model](/intangible-mcp-agents/choosing-a-model.md)
* [Troubleshooting and FAQ](/intangible-mcp-agents/troubleshooting-and-faq.md)

*Last updated 23 October 2026. Steps may change.*


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://help.intangible.ai/intangible-mcp-agents/google-gemini.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.