@microservices-sh/mcp
MCP server for microservices.sh that enables planning, inspecting, composing, validating, and deploying microservices modules and templates with preview deployments.
README
@microservices-sh/mcp
MCP server for AI agents that generate, verify, and deploy Cloudflare-native apps from reusable microservices.sh modules.
This package is a standalone stdio MCP server. It vendors a small microservices.sh SDK/module-contract snapshot until the public SDK is split out.
Why Agents Use It
@microservices-sh/mcp gives coding agents a typed planning layer for building production-oriented Cloudflare apps. Agents can discover verified templates and modules, inspect module contracts, compose an app plan, validate configuration, generate project files for review, run readiness checks, and prepare approval-gated preview deployments through the microservices.sh control plane.
It is useful when an agent needs to build apps with common product modules such as auth, booking, customer records, payments, files, and audit logs without inventing each integration from scratch.
Install
pnpm add -D @microservices-sh/mcp
Or run it directly after publish:
pnpm dlx @microservices-sh/mcp
Or install from Glama:
MCP Client Config
{
"mcpServers": {
"microservices": {
"command": "microservices-mcp",
"env": {
"MICROSERVICES_API_URL": "https://api.microservices.sh",
"MICROSERVICES_API_KEY": "favored-secret-manager-reference"
}
}
}
}
For local development inside this repo:
{
"mcpServers": {
"microservices-local": {
"command": "node",
"args": ["/absolute/path/to/mcp/dist/index.js"]
}
}
}
Build the local package first:
pnpm build
For agent-oriented one-click install guidance, see llms-install.md.
Tools
Read-only and local planning tools:
list_templatesinspect_templatelist_modulesinspect_modulelist_module_docsget_module_doccompose_appvalidate_configgenerate_projectrun_checksplan_add_modulecheck_updatesplan_module_upgradeget_secrets_statuscreate_preview_plan
Remote control-plane tools:
deploy_previewget_deployment_status
deploy_preview is mutating and requires confirm: "preview". Run create_preview_plan first.
Environment
| Variable | Purpose |
|---|---|
MICROSERVICES_API_URL |
Remote control-plane URL. Defaults to https://api.microservices.sh. |
MICROSERVICES_API_KEY |
Bearer token for remote tools. |
MICROSERVICES_TOKEN |
Fallback bearer token. |
Secret values are never returned by the MCP tools.
Registry Notes
The package is published to npm, listed in the official MCP Registry, and verified by Glama:
- npm:
@microservices-sh/mcp - MCP Registry:
sh.microservices/mcp - Glama:
@microservices-sh/mcp - awesome-mcp-servers: PR #8249
- Remote endpoint:
https://api.microservices.sh/mcp
The package declares "mcpName": "sh.microservices/mcp" in package.json. Docker/OCI packaging should wrap this same stdio server rather than forking behavior.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。