super-mcp
A gateway that aggregates multiple MCP servers into a single endpoint, namespacing their tools and forwarding calls, so an agent connects to one MCP to access the entire stack.
README
super-mcp
Multiple MCP servers put together as one. A gateway that spawns each configured sub-MCP, namespaces and re-exposes all of their tools, and forwards calls — so the agent connects to one MCP and gets the whole stack.
┌────────────── super-mcp (gateway) ──────────────┐
Claude ── stdio ──▶ │ tools/list ─▶ cloudflare__list_zones │
│ owndomain__own_domain_fanout │
│ godaddy__list_domains ... │
│ tools/call ─▶ route by prefix ─▶ sub-MCP │
└───────┬───────────────┬───────────────┬──────────┘
stdio │ stdio │ stdio │
cloudflare-mcp own-domain-mcp godaddy-mcp
└─────────────── each → its vendor REST API
Why this (and why MCP → API all the way down)
Every sub-server is itself an MCP wrapping a direct vendor REST API (Cloudflare, GoDaddy, …). The gateway changes nothing about that contract — it just unifies the surface. So the full path stays MCP → API: deterministic, least-privilege, no browser. The namespace prefix (cloudflare__, owndomain__, godaddy__) keeps it explicit which underlying API served any call, even through the aggregator.
Benefits:
- One connection, whole stack. Register
super-mcponce instead of N servers. - Add by config. Drop a server into
config/servers.json(orSUPER_MCP_SERVERS); no code change. - Decoupled. Sub-servers run as their own processes with their own scoped creds; a crash in one doesn't take down the rest (it's skipped with a logged error).
Config
config/servers.json:
{
"servers": [
{ "name": "cloudflare", "command": "node", "args": ["/path/to/cloudflare-mcp/dist/index.js"] },
{ "name": "owndomain", "command": "node", "args": ["/path/to/own-domain-mcp/dist/index.js"] },
{ "name": "godaddy", "command": "node", "args": ["/path/to/godaddy-mcp/dist/index.js"] }
]
}
Override with SUPER_MCP_SERVERS (a path to a JSON file, or inline JSON). Per-server env is merged over the parent environment.
Install
npm install && npm run build
# build the sub-servers too (they must have dist/ outputs)
claude mcp add super -- node ~/code/personal/super-mcp/dist/index.js
npm test
Notes
- Tools are advertised with their original JSON Schema passed straight through — no re-validation, no lossy Zod round-trip.
- The gateway is a low-level MCP
Server; it answerstools/listandtools/callonly. Sub-server prompts/resources are intentionally not aggregated (tools are the surface families share).
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。