gcp-mcp

GCP-MCP: Google Cloud Platform Model Control Protocol Server

⚠️ EXPERIMENTAL PROJECT ⚠️

This project provides a Model Control Protocol (MCP) server for interacting with Google Cloud Platform resources.

Overview

GCP-MCP allows AI assistants and other MCP clients to interact with Google Cloud Platform services through a standardized interface.

Installation

Prerequisites

• uv
• Access to Google Cloud Platform with appropriate permissions
• Google Cloud SDK (for authentication)

Setup

1. Clone this repository:

   git clone https://github.com/manuelbernhardt/gcp-mcp.git
   cd gcp-mcp
   

2. Install dependencies using uv:

   uv sync
   

3. Authenticate with Google Cloud:

   gcloud auth application-default login
   

Available Tools

Secret Manager Tools

The server provides the following tools for interacting with Google Cloud Secret Manager:

list_secrets

Lists all secrets in a specified GCP project.

Parameters:

• project_id (string, required): The ID of the GCP project to list secrets from
• prefix (string, optional): Filter secrets by prefix

Example:

list_secrets(project_id="my-gcp-project", prefix="api-")

delete_secret

Deletes a secret from Google Cloud Secret Manager.

Parameters:

• secret_name (string, required): The name of the secret to delete
• project_id (string, required): The ID of the GCP project containing the secret

Example:

delete_secret(secret_name="my-api-key", project_id="my-gcp-project")

add_secret

Adds a new secret or a new version to an existing secret in Google Cloud Secret Manager.

Parameters:

• secret_name (string, required): The name of the secret to add or update
• project_id (string, required): The ID of the GCP project
• secret_value (string, required): The value of the secret

Example:

add_secret(secret_name="my-new-api-key", project_id="my-gcp-project", secret_value="supersecretvalue")

Cloud Run Tools

The server provides the following tools for interacting with Google Cloud Run:

list_cloud_run_services

Lists all Cloud Run services in a specified GCP project and region.

Parameters:

• project_id (string, required): The ID of the GCP project
• region (string, required): The region where services are deployed (e.g., "us-central1", "europe-west3")

Example:

list_cloud_run_services(project_id="my-gcp-project", region="us-central1")

Response:

[
  {
    "name": "my-service",
    "uri": "https://my-service-xyz123-uc.a.run.app"
  },
  ...
]

delete_cloud_run_service

Deletes a Cloud Run service from a specified GCP project and region.

Parameters:

• service_name (string, required): The name of the service to delete
• project_id (string, required): The ID of the GCP project
• region (string, required): The region where the service is deployed

Example:

delete_cloud_run_service(service_name="my-service", project_id="my-gcp-project", region="us-central1")

Response:

{
  "status": "success",
  "message": "Cloud Run service 'my-service' successfully deleted from project 'my-gcp-project' in region 'us-central1'"
}

Configuring for Cursor

Create a mcp.json file in ~/.cursor/mcp.json:

{
  "mcpServers": {
    "gcp-tools": {
      "command": "uv",
      "args": ["--directory", "/path/to/gcp-mcp", "run", "gcp.py"]
    }
  }
}

Testing

Running Tests

# Run all tests
python -m unittest test_gcp.py

# Run specific test class
python -m unittest test_gcp.TestSecretManagerFunctions
python -m unittest test_gcp.TestCloudRunFunctions

# Run a specific test
python -m unittest test_gcp.TestSecretManagerFunctions.test_list_secrets