Stars Cash Flow MCP Server
Enables AI agents to interact with the Stars Cash Flow reseller API for real Telegram actions, including listing services, checking balance, creating orders with cost confirmation, and managing orders.
README
Stars Cash Flow — MCP server, AI skill & Hermes tool
Drive the Stars Cash Flow reseller API from any AI agent. Stars Cash Flow is a CPA exchange for real Telegram actions — channel subscribers, boosts and bot starts fulfilled by verified real users, not bots. The API is JustAnotherPanel / Perfect-Panel compatible, so it drops into any existing SMM-panel integration.
One API contract, three first-class surfaces:
| Surface | Path | For |
|---|---|---|
| MCP server (TypeScript) | src/ → dist/ |
Claude Desktop, Claude Code, any MCP host |
| Universal skill | skill/SKILL.md · openapi.yaml |
Claude Code skills, ChatGPT Custom GPT Actions |
| Hermes skill + CLI | hermes/ |
Hermes Agent, or any shell |
Documentation
- 📘 API reference — every action, field and guarantee
- 🚀 Quickstart — set up each surface in minutes
- ❓ FAQ — common questions
- ⚠️ Error reference — every error message and fix
- 🧪 Examples — runnable curl / Python / Node
The API in 30 seconds
POST https://api-stars.ros.media/api/v2 — form-encoded, dispatched on action:
| action | fields | returns |
|---|---|---|
services |
— | catalog (id, name, USD rate /1000, min, max) |
balance |
key |
{ balance, currency } |
add |
key, service, link, quantity |
{ order } — spends balance |
status |
key, order / orders |
{ charge, start_count, status, remains, currency } |
cancel |
key, orders |
{ canceled: [ids] } — refunds remainder |
Get an API key from @StarsCashFlowbot →
Reseller. Everything except services needs the key; the key holds a USD
balance. Rate limit: 60 req/min/key.
🔴 Money safety
add debits real USD. Every surface here prices the order first and refuses to
spend without explicit confirmation:
- MCP —
create_orderis two-step: the first call returns a cost estimate; it only places the order when called again withconfirm: true. - CLI —
order/cancelprint an estimate and do nothing unless--confirm. - Skill — instructs the agent to price, show the cost, and get a "yes" first.
Never hardcode or print the key — read it from STARS_CASH_FLOW_API_KEY.
Install — MCP server
Once published:
npx -y stars-cash-flow-mcp
Or from source:
git clone https://github.com/govRosMedia/stars-cash-flow-mcp.git
cd stars-cash-flow-mcp
npm install && npm run build
Register in an MCP host (claude_desktop_config.json / .mcp.json):
{
"mcpServers": {
"stars-cash-flow": {
"command": "npx",
"args": ["-y", "stars-cash-flow-mcp"],
"env": {
"STARS_CASH_FLOW_API_KEY": "your-key",
"STARS_CASH_FLOW_API_BASE": "https://api-stars.ros.media/api/v2"
}
}
}
}
Tools: list_services, get_balance, create_order (confirm-gated),
order_status, cancel_order.
See the Quickstart for ChatGPT and Hermes setup.
Use the typed client directly
import { StarsCashFlowClient } from "stars-cash-flow-mcp";
const client = new StarsCashFlowClient(); // reads STARS_CASH_FLOW_API_KEY
const services = await client.listServices();
const cost = StarsCashFlowClient.cost(services[0], 1000); // USD
Development
npm install
npm run build # tsc → dist/
npm test # vitest
npm run typecheck # tsc --noEmit
CI runs build + tests on Node 18/20/22. See CONTRIBUTING.md.
License
MIT © Stars Cash Flow
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。