GCP Infrastructure

You deploy and manage Google Cloud Platform infrastructure using natural language through an MCP server. This server translates your French requests into actions like creating VMs, handling SSH keys, running commands over SSH, and managing Terraform workflows, making cloud operations more accessible and test-friendly while you stay in a controlled development environment.

How to use

You interact with the MCP server through a client that supports the MCP standard. Start by starting the local server or connecting to a remote MCP endpoint configured for HTTPS. Use natural language queries to perform common cloud tasks: create VMs, manage SSH keys, run commands on instances, transfer files, and drive Terraform plans and applies. Your client sends a structured request to the MCP server, which executes the corresponding actions and returns results or statuses. Remember to operate in a development or isolated environment if you are testing, and prefer a domain with TLS for secure access.

How to install

Prerequisites: Install Python 3.8 or newer and ensure you have a Google Cloud project with a service account key file ready.

Step by step install and run:

1. Install Python dependencies

pip3 install -r requirements.txt

2. Place your Google Cloud service account key file in the project directory and ensure it has the necessary permissions (Compute Engine Admin and Service Account User).

3. (Optional) Install Terraform if you plan to use Terraform features

# Ubuntu/Debian
wget -O- https://apt.releases.hashicorp.com/gpg | sudo gpg --dearmor -o /usr/share/keyrings/hashicorp-archive-keyring.gpg
echo "deb [signed-by=/usr/share/keyrings/hashicorp-archive-keyring.gpg] https://apt.releases.hashicorp.com $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/hashicorp.list
sudo apt update && sudo apt install terraform

Additional sections

Configuration and security notes are important for safe development and testing. Use HTTPS with a domain and SSL certificates to enable secure access, and consider a reverse proxy to manage TLS termination. The MCP server can run behind HTTPS proxying on port 5001 by default, and you should start it in a development environment before exposing it publicly.

Security and best practices: Do not expose your MCP server publicly without authentication and TLS. Enable a firewall, implement rate limiting, and maintain audit logging. Use minimum GCP permissions for the service account and restrict access to only the resources required for your tasks.

Troubleshooting tips: Ensure dependencies are installed (pip3 install -r requirements.txt), verify the service account key file exists and is valid, and confirm the server starts correctly. If you encounter Terraform issues, verify that Terraform is installed and that your working directory contains valid Terraform configurations.

Available tools

ssh_generate_key

Generates a new SSH key pair for secure access to VMs.

ssh_list_keys

Lists all SSH keys currently stored and available for usage.

ssh_add_key

Adds an existing SSH key to the MCP server's key store.

gcp_list_instances

Lists all VM instances in your GCP project.

gcp_create_instance

Creates a new VM instance with customizable parameters and SSH key association.

gcp_start_instance

Starts a stopped VM instance.

gcp_stop_instance

Stops a running VM instance.

gcp_delete_instance

Deletes a VM instance and its resources.

gcp_get_instance

Retrieves detailed information for a specific VM instance.

ssh_execute

Executes a command on a remote VM via SSH.

ssh_upload_file

Uploads a file to a remote VM via SFTP.

terraform_init

Initializes a Terraform working directory.

terraform_plan

Plans infrastructure changes with Terraform.

terraform_apply

Applies Terraform changes to deploy infrastructure.

terraform_destroy

Destroys Terraform-managed infrastructure.

gcp_natural_query

Interprets a natural language query and maps it to cloud actions.