gcp

This MCP server implements Google Cloud platform services using Go and Cobra. It exposes MCP endpoints over the stdio transport (and can also operate via SSE as described in the project notes), providing access to Google Cloud projects, container clusters, and Cloud Run services. The server exposes operations for Projects (list and describe), Container (Clusters list and describe), and Cloud Run (Services list and describe), enabling clients to query and interact with GCP resources through the MCP protocol. When running locally, you can connect with the MCP introspector to explore the available endpoints, schemas, and capabilities, making it easy to test commands and responses from Claude Desktop or other MCP tools.

To use the server, start it with the desired transport (stdio in the example). Clients can then initiate MCP requests for Projects, Clusters, and Cloud Run services according to the MCP protocol. The inspector workflow provided in the README allows you to run a local inspector binary or use npx commands to inspect and test the server’s MCP endpoints, which is helpful for development and debugging.

Prerequisites:

• Go installed and configured (Go tooling installed).
• Git to clone or fetch the repository.
• Optional: Goreleaser for building and releasing the binary.

Install steps:

1. Ensure Go is installed: go version
2. Clone the repository or download the prebuilt binary release.
3. Build locally (example from the project): goreleaser build --snapshot --clean goreleaser release --skip=publish --snapshot --clean
4. Run the server binary with the desired transport. For stdio as shown in the example: /path/to/gcp-mcp-server --transport stdio

If you plan to deploy to Google Cloud Run, follow the deployment steps in the README (enable APIs, create service account, store credentials as a secret, and deploy with the proper environment variables such as BASE_URL and GOOGLE_APPLICATION_CREDENTIALS).

Environment integration notes:

• BASE_URL may be required to construct service endpoints for clients.
• GOOGLE_APPLICATION_CREDENTIALS should point to the service account key file (secret-managed in Cloud Secret Manager).
• Other environment variables can be added as needed for your deployment (e.g., region, project ID).

Tips and common issues:

• Ensure Google Cloud APIs used by the server (secretmanager, cloudbuild, artifactregistry, run) are enabled before deployment.
• When running locally, the inspector tool can help you discover available endpoints and payload schemas for Projects, Clusters, and Cloud Run services.
• If you switch transports (stdio vs SSE), adjust clients accordingly; the server supports both transport methods.
• For Cloud Run deployment, securely manage credentials via Secret Manager and set GOOGLE_APPLICATION_CREDENTIALS to the secret path; also configure BASE_URL to the deployed service URL.
• If you encounter authentication errors, verify the service account permissions (roles/editor is used in the provided deployment script) and ensure the service account key is accessible to the server as configured in your environment.