广告合作
MCP.so
登录

Build and Deploy a remote MCP server to Cloud Run in 5 minutes

@jackwotherspoon

代码仓库

基本信息

分类

开发工具

许可证

Apache-2.0 license

传输方式

stdio

作者

jackwotherspoon

关于 Build and Deploy a remote MCP server to Cloud Run in 5 minutes

Deploy and host a remote MCP server on Cloud Run

#mcp#cloud-run#remote-mcp-server

Build and Deploy a remote MCP server to Cloud Run in 5 minutes 配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "mcp-on-cloudrun": {
      "command": "uv",
      "args": [
        "run",
        "test_server.py"
      ]
    }
  }
}

Build and Deploy a remote MCP server to Cloud Run in 5 minutes 工具

工具列表暂未提供

当前收录暂时没有可展示的工具列表，不代表服务没有工具。请查阅项目文档了解支持的能力。

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is ☁️ 🚀 Build and Deploy a remote MCP server to Cloud Run in 5 minutes?

This project provides a quick-start template and guide to build a remote MCP server using FastMCP and deploy it to Google Cloud Run. It demonstrates how to run a Model Context Protocol server over the streamable-http (or sse) transport, making it centrally accessible, scalable, and secured with authentication. It is aimed at developers who want to move from local MCP servers to a remote, shareable infrastructure.

How to use ☁️ 🚀 Build and Deploy a remote MCP server to Cloud Run in 5 minutes?

Clone the repository, set your Google Cloud project, and deploy the server with gcloud run deploy mcp-server --no-allow-unauthenticated --region=us-central1 --source .. After deployment, run gcloud run services proxy mcp-server --region=us-central1 to create an authenticated local tunnel. Test the connection using uv run test_server.py, which exercises the add and subtract tools.

Key features of ☁️ 🚀 Build and Deploy a remote MCP server to Cloud Run in 5 minutes

  • Deploys a remote MCP server on Cloud Run with automatic scaling.
  • Enforces authentication via the --no-allow-unauthenticated flag and IAM roles.
  • Uses FastMCP for a fast, Pythonic server implementation.
  • Supports streamable-http transport (default) with sse as an option.
  • Includes two example math tools: add and subtract.
  • Provides a local proxy tunnel for secure client connections.

Use cases of ☁️ 🚀 Build and Deploy a remote MCP server to Cloud Run in 5 minutes

  • Share a centralized MCP server across a team so everyone uses the same tools.
  • Run an MCP server remotely to eliminate local setup and maintenance for each developer.
  • Leverage Cloud Run’s scalability to handle variable request loads from AI assistants.
  • Secure an MCP server with authenticated requests to prevent public access.

常见问题

1

What transports does this MCP server support?

It uses `streamable-http` by default, but you can switch to `sse` by changing the transport parameter in `server.py` to `transport="sse"`.

2

How do I deploy the MCP server to Cloud Run?

You can deploy from source with `gcloud run deploy mcp-server --no-allow-unauthenticated --region=us-central1 --source .` or from a container image built via Cloud Build. Both options require the `--no-allow-unauthenticated` flag.

3

How do I authenticate MCP clients?

Use the Cloud Run proxy by running `gcloud run services proxy mcp-server --region=us-central1`. This creates an authenticated tunnel to your service. You must also ensure that the caller (or the proxy) has the `roles/run.invoker` IAM role.

4

What are the prerequisites to get started?

You need Python 3.10+ (required by FastMCP), the Google Cloud SDK (`gcloud`), and Git for cloning the repository.

5

What tools are included in the example server?

The example Math MCP server provides two tools: `add` (returns the sum of two numbers) and `subtract` (returns the difference).

评论

开发工具 分类下的更多 MCP 服务器

JiCo for Jira & Confluence – PII-redacting MCP server with Human-in-the-Loop

drkv-com

JiCo for Jira & Confluence: local MCP server for Jira Data Center - redacts personal data before the AI model, holds every write for approval

CodeSentinel

icohangar-ops

Codebase health as MCP tools: dead code, circular dependencies, coupling, architectural drift.

Total CMS

Joe Workman

Read-only Total CMS docs and site content over MCP: exact Twig signatures, field types, schema options, the CLI, release notes and comparisons. No key required.

Mcp-Warden

ermoluk

Local-first credential broker for MCP. AI coding agents (Claude Code, Cursor, Codex…) call APIs, SSH and databases through scoped presets; secrets stay in your OS keychain and never enter the agent's context.

4

ArtiBot Cloud MCP - Zhijiangyun 智匠云

bofob

ArtiBot Cloud MCP (Zhijiangyun 智匠云) — cloud MCP server for robot manipulation data & GPU workloads. 15 tools: robot fleet management, data collection & video recording, post-processing pipelines, and benchmark evaluation

1
广告合作