Compute Engine Managed MCP Extension

The Compute Engine managed MCP extension provides a comprehensive set of capabilities that let LLM agents perform a range of infrastructure management tasks including the following:

• Manage virtual machine (VM) instances.
• Retrieve information about instance group managers, instance templates, disks, snapshots, reservations and commitments.

Google and Google Cloud managed MCP servers can be used in your AI applications with enterprise-ready governance, security, and access control.

1. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

2. Get your administrator to grant you the MCP Tool User role (roles/mcp.toolUser) on the Google Cloud project. If you created a new project, then you already have the required permissions.

3. Ensure your administrator has enabled the Compute Engine API on the Google Cloud project.

This extension uses Google Application Default Credentials (ADC) to perform authentication. To login with ADC, run the following command in your terminal:

gcloud auth application-default login

For additional details, see the ADC documentation.

To install the extension, run the following command in your terminal:

gemini extensions install https://github.com/gemini-cli-extensions/compute-engine

To see a complete list of available tools and their schemas, see the Compute Engine MCP reference.

The following sample use cases describe how you can use the Compute Engine MCP server to manage Compute Engine resources:

• Inspect and manage resources. For example, to understand resource allocation and configuration in your project, you can list all compute instances. You can also find all running compute instances in a zone that have a specific accelerator attached, and show their location and name for resource management.
• Clean up unused resources to reduce operational costs. For example, identify disk snapshots in a zone that are no longer associated with a source disk, or identify and delete stopped VM instances that have costly GPU resources attached.
• Optimize instance performance. For example, resize an under-provisioned VM instance to a larger machine type in the same family, and confirm the successful update.
• Provision specialized VMs for AI workloads with zone flexibility. For example, create a VM instance with a specific GPU accelerator attached, in any zone in a specified region where it is available.
• Troubleshoot and validate instance configurations. For example, retrieve configuration details for a specific VM instance where the job is frozen, reboot it, and confirm the underlying accelerator and disk are attached.

The following are sample prompts that you can use to perform tasks by using the Compute Engine MCP server:

• List all VMs in PROJECT_ID, including the VM name and zone.
• Show the instance details for VM_NAME.
• In REGION, find all disk snapshots for which the source disk no longer exists.
• Change the machine type of VM_NAME to the next largest machine type in the same machine family, send notification when it's back online, and confirm the new machine type.
• Find all running VMs in REGION with NVIDIA accelerators, and show the zone and name for these VMs.
• Create a VM in ZONE with an NVIDIA T4 accelerator attached. Name the VM my-nvidiat4-vm.
• Find all stopped VMs in REGION with NVIDIA Tesla T4 accelerators, and delete them.

Replace the following:

• PROJECT_ID: the Google Cloud project ID.
• REGION: the name of the region where your resources exist.
• ZONE: the name of the zone where your VMs exist.
• VM_NAME: the name of your VM instance.

MCP introduces new security risks and considerations due to the wide variety of actions that you can take with MCP tools. To minimize and manage these risks, Google Cloud offers defaults and customizable policies to control the use of MCP tools in your Google Cloud organization or project.

For more information about MCP security and governance, see AI security and safety.

The Compute Engine MCP server doesn't have its own quotas. There is no limit on the number of call that can be made to the MCP server. You are still subject to the quotas enforced by the APIs called by the MCP server tools.

• Explore the Compute Engine remote MCP server reference documentation, which includes a list of all available tools, and the full input and output schema for each tool.
• See the Compute Engine overview.
• Learn about MCP security and governance.