mdapi-mcp
MCP server for converting HTML to Markdown and extracting page data, with pay-per-call using USDC on Base.
README
mdapi-mcp — MCP server for https://md.fastdb.in
Exposes the paid mdapi endpoints as MCP tools, with x402 pay-per-call ($0.005 in USDC on Base, no signup, no API key):
| Tool | What it does | Price |
|---|---|---|
html_to_markdown |
URL or raw HTML → clean, LLM-ready Markdown. render: true executes the page in a real headless browser first (JS-rendered SPAs). |
$0.005 |
extract_page |
URL or raw HTML → structured JSON: title, byline, excerpt, links, images, meta. | $0.005 |
service_info |
Service + pricing details. | free |
Failed calls are never charged — settlement happens only after the tool succeeds.
Two ways to use it
1. Remote MCP server (recommended)
Streamable HTTP endpoint: https://md.fastdb.in/mcp
Listed in the official MCP registry as in.fastdb.md/mdapi.
Payments are in-protocol x402 v2 (the official @x402/mcp pattern): an unpaid
tools/call returns the payment requirements; an x402-aware MCP client signs a
USDC transferWithAuthorization and retries with the payment in _meta.
No ETH needed — the facilitator sponsors gas.
import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js";
import { createx402MCPClient } from "@x402/mcp";
import { ExactEvmScheme } from "@x402/evm/exact/client";
import { privateKeyToAccount } from "viem/accounts";
const client = createx402MCPClient({
name: "my-agent", version: "1.0.0",
schemes: [{ network: "eip155:8453", client: new ExactEvmScheme(privateKeyToAccount(process.env.EVM_PRIVATE_KEY)) }],
autoPayment: true,
});
await client.connect(new StreamableHTTPClientTransport(new URL("https://md.fastdb.in/mcp")));
const result = await client.callTool("html_to_markdown", { url: "https://example.com" });
server.js is this remote server. It runs conversion in-process (same ../lib
as the HTTP API) and verifies/settles via the Coinbase CDP facilitator.
Deployed as systemd unit biz-mdapi-mcp on 127.0.0.1:8403 behind nginx /mcp.
2. Local stdio server (Claude Desktop and other vanilla MCP clients)
stdio.js runs on your machine with your wallet key and wraps the live paid
HTTP API: it answers the HTTP 402 challenge for you (x402 v1 X-PAYMENT flow
via x402-fetch). Install straight from the tarball we host:
// Claude Desktop / any MCP client config
{
"mcpServers": {
"mdapi": {
"command": "npx",
"args": ["-y", "https://md.fastdb.in/mcp/mdapi-mcp.tgz"],
"env": { "EVM_PRIVATE_KEY": "0x..." } // wallet holding USDC on Base
}
}
}
Env: EVM_PRIVATE_KEY (required), MDAPI_BASE_URL (optional override).
Per-call spend is hard-capped at 0.01 USDC in the client wrapper.
Discovery
- Official MCP registry:
in.fastdb.md/mdapi(remote, streamable-http) - x402 Bazaar (Coinbase CDP discovery layer):
https://md.fastdb.in/mdand/extractare indexed; agents using the Bazaar MCP server (https://api.cdp.coinbase.com/platform/v2/x402/discovery/mcp) can find and call them there too. - Machine-readable catalog:
GET https://md.fastdb.in/(JSON for non-browsers)
Repo layout
server.js— remote MCP server (x402 v2 in-protocol payments, CDP facilitator). Reference: it imports mdapi's core../lib, which lives in the main mdapi codebase and is not vendored here —stdio.jsis the fully self-contained entry.stdio.js— local stdio MCP server (pays the HTTP API with your key)server.json— MCP registry manifest
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。