marketifyall-mcp
A thin bridge that connects any stdio-based MCP client to the MarketifyALL workspace, enabling content generation, search, and discovery through a hosted MCP server with graduated safety controls.
README
marketifyall-mcp
Stdio ⇄ Streamable-HTTP bridge that connects any local MCP client (Claude Code, Codex CLI, Gemini CLI, …) to a MarketifyALL workspace.
MarketifyALL exposes its capability layer as a hosted MCP server at https://app.marketifyall.com/api/mcp (Streamable HTTP, bearer-token auth). Clients with native remote-MCP support can connect to that URL directly — this bridge is the universal-compatibility fallback for clients that only speak stdio.
How it works
┌────────────────┐ stdio ┌───────────────┐ HTTPS (Streamable HTTP) ┌────────────────────────────┐
│ MCP client │ ─────────► │ this bridge │ ───────────────────────────► │ app.marketifyall.com/api/ │
│ (Claude Code, │ ◄───────── │ (~40 LOC) │ Authorization: Bearer … │ mcp — capability MCP server│
│ Codex, …) │ └───────────────┘ └────────────────────────────┘
└────────────────┘
The bridge is deliberately thin: it opens one upstream connection to the hosted server and proxies tools/list and tools/call through 1:1. No tool logic lives here — every capability, schema, and permission decision stays server-side, so the bridge never goes stale when the workspace gains new tools.
On the hosted side, tools are generated from a scope-filtered capability registry (Zod schemas become MCP tool schemas at registration time). Clients see a small advertised tier plus two discovery tools — search_capabilities and invoke_capability — so the long tail of capabilities is discoverable on demand instead of flooding the client's context with every tool definition up front.
Safety model
MCP clients can act with real consequences (posting content, sending email), so the hosted server enforces a graduated model:
- Reads are free — listing content, searching, fetching drafts.
- Generation costs workspace credits — drafting posts, emails, SEO metadata.
- Publishing and sending are never executed directly — those calls create a pending action that a human approves in the MarketifyALL Command Center. The agent proposes; the user disposes.
Usage
Requires Node.js 18+ and a MarketifyALL account.
- Create a token in MarketifyALL → Settings → MCP Tokens.
- Clone and run:
git clone https://github.com/Aadilhassan/marketifyall-mcp.git
cd marketifyall-mcp
MARKETIFYALL_TOKEN=mk_live_... npx -y tsx index.ts
- Or register it with your MCP client, e.g. in
.mcp.json:
{
"mcpServers": {
"marketifyall": {
"command": "npx",
"args": ["-y", "tsx", "/path/to/marketifyall-mcp/index.ts"],
"env": { "MARKETIFYALL_TOKEN": "mk_live_..." }
}
}
}
Configuration
| Env var | Required | Description |
|---|---|---|
MARKETIFYALL_TOKEN |
yes | Bearer token created in MarketifyALL settings |
MARKETIFYALL_MCP_URL |
no | Override the endpoint (e.g. http://localhost:3000/api/mcp for local development) |
Status
Extracted from the MarketifyALL monorepo. Publishing to npm as @marketifyall/mcp is planned; until then, run from source as above.
License
MIT © Aadil Hassan
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。