holdings-mcp
Let your AI send invoices and take payment — card or ACH. Free. Every write is confirm-gated, and it connects Claude, ChatGPT, or Cursor to your Holdings workspace.
README
Holdings — MCP server
Software your AI can run. Let Claude, ChatGPT, or Cursor send invoices and take payment — card or ACH. You approve every send. No monthly fee.
Tell the AI you already use who to bill and how much: it drafts the invoice, shows you a preview, and sends it the moment you confirm. Your client gets a payment link; you get paid by card or ACH, and you're told when the money lands. Quotes, clients, and recurring invoices included.
Get a key
Create a free key in your Holdings workspace → Settings → API keys. No monthly fee.
Connect
Pick whichever fits your client. All three use the same free key.
1. Local (npm / stdio) — Claude Desktop, Cursor, Cline, Windsurf
{
"mcpServers": {
"holdings": {
"command": "npx",
"args": ["-y", "@getholdings/mcp"],
"env": { "HOLDINGS_API_KEY": "hw_live_your_key_here" }
}
}
}
2. Hosted (remote / Streamable HTTP) — any client that supports remote MCP
https://mcp.getholdings.com/mcp
Authenticate with either an Authorization: Bearer <your key> header or a ?mcpKey=<your key> query parameter. Nothing to install.
3. Smithery
One-click via the registry: smithery.ai/servers/holdings/invoicing — paste your key when prompted.
Then just ask your AI: "Send Acme the December retainer — $8,400, net 30."
What you can do
- Invoices — draft, preview, send, and track; get paid by card or ACH.
- Quotes — send estimates; turn accepted quotes into invoices.
- Clients — keep the people and businesses you bill.
- Payments — see what's paid and what's outstanding.
Every action is confirmable: a draft is created first, and nothing reaches a client until an explicit send. Amounts are recomputed server-side. You stay in control.
Links
- Learn more: https://getholdings.com/invoicing
- npm: https://www.npmjs.com/package/@getholdings/mcp
- Registry: https://registry.modelcontextprotocol.io (
com.getholdings/mcp) - Privacy: https://getholdings.com/privacy · Terms: https://getholdings.com/terms
Built by Holdings Financial Technologies. Payments via Stripe.
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。