mcp-china-ecosystem
by 543366565
README.md
# MCP China Ecosystem π¨π³
> **Bridging AI Agents to the Chinese Internet.** Search Baidu, explore Bilibili, extract Chinese web content β no API keys required.
A [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server that connects AI agents (Claude Desktop, Cursor, Hermes Agent, etc.) to China's internet ecosystem.
## Tools
| Tool | Description |
|------|-------------|
| **`baidu_search_tool`** | Search Baidu β China's #1 search engine. Returns titles, URLs, and snippets in Chinese. |
| **`bilibili_search_tool`** | Search Bilibili β China's largest video platform. Returns videos with creator, views, danmaku, likes. Supports sorting by relevance, popularity, date, or danmaku count. |
| **`extract_chinese_page_tool`** | Extract clean text content from Chinese websites (Zhihu, Bilibili, 36Kr, CSDN, Jianshu, etc.). Strips ads and navigation noise. |
## Quick Start
```bash
pip install mcp-china-ecosystem
```
Or from source:
```bash
git clone https://github.com/543366565/mcp-china-ecosystem.git
cd mcp-china-ecosystem
pip install -e .
```
### Run
```bash
# stdio mode (for Claude Desktop, Cursor, Hermes)
mcp-china-ecosystem
# SSE mode (remote access)
mcp-china-ecosystem --transport sse --host 0.0.0.0 --port 8000
```
### Configure Claude Desktop
Edit `claude_desktop_config.json`:
```json
{
"mcpServers": {
"china-ecosystem": {
"command": "mcp-china-ecosystem"
}
}
}
```
## Features
- **Zero API keys** β All services use public interfaces (Bilibili open API, Baidu web scraping)
- **Chinese-optimized** β Built for Chinese-language content and Chinese website structures
- **Lightweight** β Single dependency (parsel + httpx), no heavy frameworks
- **Fast** β Async HTTPX for concurrent requests
## Why This Matters
AI agents are great at searching the global web, but **China's internet is a walled garden**. Baidu, Bilibili, Zhihu, and other Chinese platforms are inaccessible to most MCP tools. This server fills that gap β giving AI agents direct access to the Chinese internet.
## License
MIT
TDQS
A3.8/5.0
Scored across 3 tools
Disambiguation5/5
Each tool targets a distinctly different function: web search, video search, and page extraction. No ambiguity in purpose.
Naming Consistency5/5
All tools follow a consistent verb_noun pattern with snake_case (baidu_search_tool, bilibili_search_tool, extract_chinese_page_tool).
Tool Count4/5
Three tools is minimal but covers essential operations for the China ecosystem domain. Slightly thin but not unreasonably so.
Completeness3/5
Missing tools for other major platforms (Weibo, Zhihu, etc.) and common operations like translation or user info. Gaps exist but core search and extraction are present.
Maintenance
ActivityStale
ResponsivenessNo issues