Cloud Run MCP server
Google's MCP server for deploying apps to Cloud Run from an agent, with service listing and log reading
Install
"cloud-run": {
"command": "npx",
"args": ["-y", "@google-cloud/cloud-run-mcp"],
"env": {
"GOOGLE_CLOUD_PROJECT": "PROJECT_NAME",
"GOOGLE_CLOUD_REGION": "PROJECT_REGION"
}
}Run gcloud auth login and gcloud auth application-default login first. The env block is optional.
This is third-party code. Review the repository files before installing.
What it does
The server deploys code to Cloud Run, either from file contents passed in or, when run locally, from a folder on disk. The agent can list services in a project and region, get service details and read logs with errors. Locally it can also list projects and create a new project linked to a billing account. The deploy and logs prompts deploy the current folder or show logs in one step. The repository also ships gcloud-based Cloud Run skills.
Who it is for. For developers who want the agent to ship apps to Cloud Run itself.
Good fit when
- You want to ship a prototype to Cloud Run straight from an IDE or CLI agent
- You need logs and errors of a deployed service
- You need a new GCP project for an experiment
Not a fit when
- Deploys go through reviewed CI and the agent should not ship directly
- You do not host on Google Cloud
Example request
Deploy the current folder to Cloud Run as service hello-world in europe-west1Limitations
Requires the gcloud CLI, gcloud auth login and application-default login, and a billed Google Cloud account. SKIP_IAM_CHECK defaults to true, so deployed services become public. The authors warn against running the remote mode without IAM authentication. Google Cloud does not accept sign-ups or payments from Russia.
How to disable. Remove the cloud-run block from your MCP client config or run gemini extensions uninstall cloud-run.
MCP
- Transport
- stdio, sse, http
- Authentication
- OAuth
| Environment variables | |
|---|---|
| GOOGLE_CLOUD_PROJECT | Default project ID |
| GOOGLE_CLOUD_REGION | Default region |
| DEFAULT_SERVICE_NAME | Default service name |
| SKIP_IAM_CHECK | Defaults to true, making services public. false enables IAM checks |
| GOOGLE_APPLICATION_CREDENTIALS secret | Path to the credentials file when running in Docker |
Security check
- Deploys code to a Google Cloud production environment
- Can create projects linked to a billing account
- Makes services public by default
README in short
The README lists the server's tools and prompts, environment variables and three ways to run it: npx, Docker, and remotely on Cloud Run behind an IAM proxy. It also covers installing as a Gemini CLI extension and an OAuth mode. At the end it introduces Cloud Run skills installed with gemini skills install.
FAQ
How does local mode differ from remote?
Locally the server uses your gcloud credentials and can deploy folders and create projects. Remotely it runs on Cloud Run behind an IAM proxy and deploys only to its own project.
Is this about hosting MCP servers on Cloud Run?
No, this server deploys your code. Google has separate docs on hosting MCP servers on Cloud Run.
Related
An MCP server built into the Netdata agent: metrics, logs, alerts and live process, service and container data for an AI assistant
GitHub's official MCP server: code, issues, pull requests, Actions and security alerts straight from the agent
AWS's official MCP server suite: docs, IaC, containers, serverless, databases, cost and monitoring
A GitHub Action that runs Claude Code in pull requests and issues: answers mentions, reviews and makes changes