Provides an MCP server to manage Google Cloud Platform resources via an MCP client for compute, storage, networks, and more.
Configuration
View docs{
"mcpServers": {
"henihaddad-gcp-mcp": {
"command": "uvx",
"args": [
"gcp-mcp"
]
}
}
}You can manage Google Cloud Platform resources through an MCP client by running a lightweight MCP server that communicates with the client, enabling natural language style commands to interact with GCP services like compute, storage, networks, and databases.
After you set up the MCP client to connect to the GCP MCP Server, you can ask it to perform common tasks such as listing projects, starting or stopping compute instances, or creating resources like storage buckets. Use natural language to request actions, for example: list my GCP projects, show my compute instances, or create a new storage bucket. When actions require authentication, you will be prompted to authenticate in a browser window; complete the authentication to grant access. The MCP server will execute the requested operations through the connected client and report back results.
Prerequisites: ensure you have Python installed on your machine.
Install the GCP MCP package using Python's package manager.
Configure your MCP client to connect to the GCP MCP Server by adding the following stdio configuration (this enables local execution via the uvx runtime). The configuration must be provided to your MCP client as shown in the example.
{
"type": "stdio",
"name": "gcp_mcp",
"command": "uvx",
"args": [
"gcp-mcp"
]
}Configuration shows the MCP server endpoint you will run locally. Use the stdio configuration above to wire your client to the local MCP server. When you start a session and issue commands, you may see a browser authentication prompt if your GCP credentials require it.
Security tip: limit access to the MCP client and ensure you revoke access if credentials are compromised. Keep your authentication flow within trusted environments.