> For the complete documentation index, see [llms.txt](https://docs.datasentinel.io/manual/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.datasentinel.io/manual/ai-connector/connect-a-desktop-app/configure-gemini-desktop-app.md). # Configure the Google Gemini desktop app Use Datasentinel from the Google Gemini desktop app: add the MCP server as a custom app, sign in, and upload the Datasentinel Skill. This page shows how to reach your Datasentinel platform from the **Gemini desktop app**, Google's application for Windows. The setup is not done in the desktop app. Google attaches MCP servers and Skills to your **Google account**, and you add both in the Gemini web app, at [gemini.google.com](https://gemini.google.com). Gemini calls an MCP server added this way a **custom app**. {% hint style="info" %} The steps below follow Google's documentation, last checked on 6 October 2026. Google changes the Gemini app often. {% endhint %} {% hint style="warning" %} **Network requirements.** Gemini calls a custom app from Google's servers, never from your workstation, even when you type in the desktop app. Access from your workstation or its VPN is not sufficient. Google's servers must reach the MCP endpoint, `https:///mcp`, and the sign-in endpoints on the same host, with a publicly trusted certificate. A platform on a private network does not work as-is. See [MCP servers on a private network](/manual/ai-connector/troubleshooting/private-network-mcp-servers.md). For a Google assistant that runs on your workstation, see [A private platform](#a-private-platform) below. {% endhint %} ## What works where | Where you use Gemini | Datasentinel MCP server (custom app) | Datasentinel Skill | | ---------------------- | ------------------------------------ | ---------------------- | | Gemini web app | Add, manage and use | Upload, manage and use | | Gemini app for Windows | Not listed by Google | Not listed by Google | Google states that custom apps are available only in the Gemini web app and mobile app, and Skills only in the web app, the mobile app and the Gemini app on Mac. Its documentation does not list the Windows app for either. The [test step](#test-in-the-desktop-app) below tells you what your own account gets. If the desktop app does not show the Datasentinel app, use [gemini.google.com](https://gemini.google.com) in a browser, with the same Google account. ## Before you start * A **personal Google account**, for a user aged 18 or over and located **in the United States**. Google does not offer custom apps to work or school accounts, nor outside the United States. For a work account, see [Work accounts: Gemini Enterprise](#work-accounts-gemini-enterprise). * Gemini used in English. Custom apps are available in English only. * **Keep Activity** turned on in Gemini. Custom apps and Skills are unavailable when it is off. * A Datasentinel platform in release 2026.10 or later, the first release with the OAuth sign-in, reachable as described in the network requirements above. See [MCP Server Setup](/manual/ai-connector/server-setup/mcp-server-setup.md). * A Datasentinel account for each user: the same login and password as for the web interface. Google states that Skills are available without a Google AI subscription, and lists no subscription among the requirements for custom apps. ## Install the Gemini desktop app The app needs Windows 10 or later. Download `GeminiSetup.exe` from [gemini.google/desktop](https://gemini.google/desktop), run it, then open **Gemini** from the Start menu and sign in with your Google account. `Alt + Space` opens Gemini over your active work. See Google's [Use the Gemini app for Windows](https://support.google.com/gemini/answer/18263854). ## Set it up {% stepper %} {% step %} #### Add the Datasentinel MCP server as a custom app On your computer, open [gemini.google.com](https://gemini.google.com) in a browser. Sign in with the same Google account as in the desktop app. At the bottom, select **Settings & help**, then **Connected Apps**. Under **Custom apps**, use the **Add a custom app** field, enter the MCP server URL, `https:///mcp`, and select **Next**. Leave **Advanced features** closed. It takes an OAuth client ID and secret, for servers without dynamic client registration. The Datasentinel server supports it, so the URL is all Gemini needs: no token, no header. {% hint style="warning" %} **Keep the `datasentinel-` prefix if Gemini asks for a name.** The Skill finds its MCP servers by a name starting with `datasentinel-`, for example `datasentinel-production`. If the `datasentinel platforms` check lists nothing after the setup, check that the custom app is connected and turned on: see Google's [Use & manage Connected Apps](https://support.google.com/gemini/answer/13695044). {% endhint %} See Google's [Connect & manage custom apps](https://support.google.com/gemini/answer/17209137). {% endstep %} {% step %} #### Sign in Follow the on-screen instructions. Gemini registers itself with the platform, then opens the Datasentinel login page. Sign in with your **platform login and password**, the same ones as the Datasentinel web interface. Your password goes to the login page only: Gemini never sees it and never asks for a token. Each user signs in with their own account, so the platform applies that user's access rights. Gemini then keeps the session and renews it by itself. See [How authentication works](/manual/ai-connector/server-setup/how-authentication-works.md). If Gemini asks for an OAuth client ID and secret instead of opening the login page, its automatic registration with the platform failed. Check the network requirements above: Google's servers must reach `/register` and `/token`, not only `/mcp`. See [Sign-in problems](/manual/ai-connector/troubleshooting/sign-in-problems.md), and Google's [Connect & manage custom apps](https://support.google.com/gemini/answer/17209137) for what the **Advanced features** fields expect. {% endstep %} {% step %} #### Upload the Datasentinel Skill Download [datasentinel-skill.zip](https://app.datasentinel.io/ds-api/download/datasentinel-skill.zip), the same archive as for the other assistants. Still in the Gemini web app, select **Settings** in the sidebar, then **Skills**. Select **Upload** at the top of the page, choose the archive, then **Open**. Review the Skill, and select **Create**. Google documents three accepted uploads: a `SKILL.md` file, or a folder or `.zip` archive with a `SKILL.md` file in its main folder, up to 100 MB in total. If Gemini does not find `SKILL.md` in the archive, extract it and upload the `datasentinel` folder it contains. {% hint style="warning" %} **"Only SKILL.md file is allowed".** Google makes Skills available gradually, and the upload dialog of some accounts refuses the archive and the folder with this message. Google documents the folder and `.zip` upload for the web app only, so first retry at [gemini.google.com](https://gemini.google.com) in a browser, not in the desktop app. If the web app shows the same message, extract the archive and upload the `datasentinel/SKILL.md` file alone. Gemini has no other way to add the supporting files: Google states that reference files can only be uploaded with the Skill. The Skill then runs on its main instructions, without its `references/` and `assets/` files (detailed analysis workflows, communication styles and report templates). When your account accepts the archive, replace the Skill with it: see [Updating the Skill](#good-to-know). {% endhint %} The Skill is then attached to your Google account. Gemini applies an active Skill automatically when your prompt relates to Datasentinel. To call it yourself, type `/` in the prompt (or `@`, if your account has Google's updated shortcut) and select `datasentinel`. See Google's [Create & manage skills](https://support.google.com/gemini/answer/17094296). {% endstep %} {% step %} #### Test in the desktop app Open the Gemini desktop app and start a new chat. Type `@` and select the Datasentinel app, so that Gemini uses it for the prompt. Then ask: | Ask Gemini for… | Expected result | | ------------------------ | -------------------------------------------------------------------------------- | | `datasentinel status` | The platform you expect, reported healthy. This confirms the MCP server answers. | | `datasentinel whoami` | Your own Datasentinel login, your privilege and the session expiry. | | `datasentinel platforms` | Every Datasentinel MCP server the Skill discovered. | The full command list is in the [Skill commands reference](/manual/ai-connector/use-the-assistant/skill-commands.md). If `@` does not list the Datasentinel app in the desktop app, run the same test at [gemini.google.com](https://gemini.google.com). When it works there, the setup is correct, and the desktop app does not offer custom apps to your account yet. Keep using the web app, with the same Google account. {% endstep %} {% endstepper %} ## Good to know * **Several platforms.** Add one custom app per Datasentinel platform, each with its own URL. You sign in to each one separately, and one Skill serves them all. * **Gemini Spark.** Custom apps also serve Gemini Spark, the agent of the Gemini app. Spark has its own availability and plan requirements: check Google's documentation. * **Read-only tools.** The Datasentinel tools only read monitoring data. If Gemini asks you to confirm a tool call, approving it is safe. * **Turn off, disconnect, remove.** In **Connected Apps**, turn the custom app off to stop Gemini using it. **More details** › **Disconnect** unlinks the MCP server and ends the sign-in. **More details** › **Remove app** deletes the custom app from your Google account. * **Updating the Skill.** Manage the Skill on the **Skills** page of the web app: deactivate it, delete it, or download it as a `.zip` archive. To update it, open the Skill, select **More** › **Replace skill**, and upload the new archive. Google requires the whole Skill each time, with all its files. Do this in the web app: Google reports that Skills with uploaded files cannot be edited in its other apps for now. ## Work accounts: Gemini Enterprise The Gemini app does not offer custom apps to work or school accounts. For an organization, Google's path is **Gemini Enterprise**, a separate product at [business.gemini.google](https://business.gemini.google), not the Gemini desktop app. In Gemini Enterprise, Business edition, a team administrator adds the MCP server under **Manage team** › **Connected apps** › **Add MCP Server**. Google documents this feature as a pre-GA offering. The network requirements are the same: a public HTTPS endpoint, with the streamable-HTTP transport, which is what the Datasentinel server serves. This path does not use dynamic client registration. It needs an OAuth client ID and secret registered in advance, with Google's redirect address. The connection brings the tools only, not the Skill. To set it up, follow Google's [Set up your custom MCP server connection](https://support.google.com/g/answer/17106276). ## A private platform The Gemini app cannot reach a platform that has no public endpoint, and Google documents no tunnel for it. Two options keep the platform private: * **Antigravity 2.0**, Google's desktop agent app. It runs Gemini models and calls the MCP server from your workstation, over your LAN or VPN. See [Use the Antigravity desktop app](/manual/ai-connector/connect-a-cli-assistant/configure-antigravity-cli.md#use-the-antigravity-desktop-app). * **A gateway** that publishes only the MCP host. See [Keep the platform private](/manual/ai-connector/troubleshooting/private-network-mcp-servers.md#keep-the-platform-private). ## Google documentation Google changes these products often. Check its current documentation: | Topic | Official documentation | | ----------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | The desktop app | [Use the Gemini app for Windows](https://support.google.com/gemini/answer/18263854), [The Gemini app is now available for Windows](https://blog.google/innovation-and-ai/products/gemini-app/gemini-app-now-on-windows/) | | Custom apps (MCP) | [Connect & manage custom apps](https://support.google.com/gemini/answer/17209137), [Use & manage Connected Apps](https://support.google.com/gemini/answer/13695044) | | Skills | [Create & manage skills](https://support.google.com/gemini/answer/17094296), [About the transition from Gems to skills](https://support.google.com/gemini/answer/18560919) | | Gemini Enterprise | [Set up your custom MCP server connection](https://support.google.com/g/answer/17106276) | --- # 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://docs.datasentinel.io/manual/ai-connector/connect-a-desktop-app/configure-gemini-desktop-app.md?ask=&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.