Chinaready MCP

Search China market-entry guides (ICP licensing, app stores, China cloud regions) with source-URL citations; topic catalog and full-guide Markdown retrieval.

Servidor MCP hospedado

npx add-mcp 'https://mcp.chinaready.co/mcp'

Instala no Claude Code, Codex, Cursor e outros

Documentação

Chinaready MCP Server

Chinaready China Market-Entry Search MCP connector – tool definition quality and endpoint health on Glama

Chinaready MCP - Let your AI cite real sources on China market entry | Product Hunt

Remote MCP server for chinaready.co — the practical guide for Western SaaS, app, and game teams entering Mainland China. Ask about ICP filing and licensing, app-store and game publishing, China cloud regions (AWS China, Azure China, Alibaba Cloud), payments, and compliance; every answer cites its source page.

  • Endpoint: https://mcp.chinaready.co/mcp (Streamable HTTP, stateless)
  • Auth: none — public, read-only content
  • Host: any Cloudflare Workers-compatible runtime (zero dependencies, zero secrets)

Tools

ToolWhat it does
searchHybrid semantic + keyword search over the corpus; returns ranked excerpts with absolute chinaready.co URLs
list_topicsThe topics the corpus covers, with example queries and hub pages
get_articleReads any chinaready.co page as clean Markdown

All tools are read-only (readOnlyHint: true).

Install

Claude — claude.ai → Settings → Connectors → Add custom connector → paste https://mcp.chinaready.co/mcp → choose No sign in → Add.

Claude Code

claude mcp add --transport http chinaready https://mcp.chinaready.co/mcp

Cursor — one click:

https://cursor.com/en/install-mcp?name=chinaready&config=eyJ1cmwiOiJodHRwczovL21jcC5jaGluYXJlYWR5LmNvL21jcCJ9

VS Code — one click (opens the redirect installer):

https://insiders.vscode.dev/redirect/mcp/install?name=chinaready&config=%7B%22url%22%3A%22https%3A%2F%2Fmcp.chinaready.co%2Fmcp%22%7D

or in .vscode/mcp.json:

{ "servers": { "chinaready": { "url": "https://mcp.chinaready.co/mcp" } } }

Any MCP client (Windsurf, Zed, LibreChat, …) — remote server, no command, no keys:

{ "mcpServers": { "chinaready": { "url": "https://mcp.chinaready.co/mcp" } } }

Development

npm install
npm test        # vitest unit tests (protocol + tools, upstream mocked)
npm run deploy  # wrangler deploy (needs your Cloudflare auth)

The server is a thin stateless proxy over Chinaready's public retrieval API (search.chinaready.co); it stores nothing and needs no environment variables. get_article only fetches chinaready.co pages it has validated as safe site paths.

Usage tracking

Every RPC message writes one data point to the chinaready_mcp_usage Analytics Engine dataset (blobs: method, tool, outcome, truncated ua; retention ~90 days). Method and tool names only — arguments and query content are never logged.

CLOUDFLARE_API_TOKEN=… CLOUDFLARE_ACCOUNT_ID=… npm run usage        # last 7 days
CLOUDFLARE_API_TOKEN=… CLOUDFLARE_ACCOUNT_ID=… npm run usage 30     # last 30 days

Prints the daily volume, the RPC funnel (initialize → tools/list → tools/call), tool-call breakdown with error outcomes, and a client split (sdk / probe-bot / browser) so real usage can be told apart from the MCP directory liveness bots. The token needs Analytics Engine Read.

Raw queries go through the Analytics Engine SQL API (POST /accounts/{account}/analytics_engine/sql):

SELECT blob2 AS tool, sum(_sample_interval) AS hits
FROM chinaready_mcp_usage
WHERE timestamp > now() - INTERVAL '7' DAY AND blob1 = 'tools/call'
GROUP BY tool ORDER BY hits DESC

License

MIT