hust-library-thesis
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@hust-library-thesisSearch CNKI for 2023 PhD theses on large language models"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
hust-library-thesis
华中科技大学图书馆(数字资源统一访问平台 libresource.hust.edu.cn)→ 中国知网(CNKI)
学位论文 的 MCP 服务。对外契约对齐 kimi-datasource:get_data_source_desc +
call_data_source_tool 两个工具,结果按 file_path 落盘、只回传预览。
Agent 用户请读 AGENT_GUIDE.md —— 面向 AI agent 的完整使用说明 (接入方式、调用配方、纪律、失败处理、排错速查);本 README 是人类视角概览。
复刻视频《访问知网阅读文献》的完整链路: 数字资源统一访问平台 → 智能访问 → 知网学位论文库 → 检索 → 详情 → PDF 下载 → 本地分页阅读。
工具面
工具 | 说明 |
| 单数据源 API 文档(markdown),先查再用 |
| 分发调用,返回 |
数据源 hust_cnki_thesis 的 7 个 API:
check_access— 通道体检(智能访问闸 + 知网机构态识别),轻量、默认无头open_login— 弹有头浏览器做一次性登录(统一身份认证 → 智能访问),后台守护进程轮询, 登录态持久化在.browser-profile/search_theses— kns8s 学位论文库检索,CSV 落盘(title/authors/school/degree/year/cited/downloads/url)get_thesis_detail— 详情结构化抓取(摘要/关键词/DOI/导师/TOC/全文按钮),md 落盘download_thesis— PDF 下载(bar.cnki.net 下单链),落盘 + manifest 登记(键=DOI), 可选extract_text抽全文 markdown;单次一篇,批量需显式授权(batch参数直接拒绝)read_thesis— 本地 PDF 分页文本读取(PyMuPDF),pages="1-8"/"52-56"/"all"list_theses— 已下载清单 CSV
Related MCP server: HUST CNKI Research MCP
目录
D:\hust-library-thesis\
├── server\hust_library_thesis_server.py # MCP 服务器(单文件,FastMCP)
├── .browser-profile\ # Playwright 持久档案(登录态/风控 cookie)
├── out\
│ ├── datasource\ # 检索/详情/清单默认落盘处
│ ├── downloads\ # PDF + manifest.json
│ └── shots\ # 风控/登录截图(人工核对用)
└── tests\test_client.py # stdio 冒烟客户端注册(ZCode)
C:\Users\<用户名>\.zcode\cli\config.json → mcp.servers:
"hust-library-thesis": {
"type": "stdio",
"command": "C:\\Users\\<用户名>\\AppData\\Local\\Programs\\Python\\Python312\\python.exe",
"args": ["D:\\hust-library-thesis\\server\\hust_library_thesis_server.py"],
"timeoutMs": 600000,
"enabled": true
}改完重启 ZCode 生效。配套 skill 在 C:\Users\<用户名>\.agents\skills\hust-library-thesis\SKILL.md。
设计要点(与 example-kit case_online 同源纪律)
浏览器走子进程:Playwright 在
--op子进程里运行,结果经文件回传——浏览器组件污染 stdout 会破坏 MCP framing(实测踩坑:管道写端被 Edge 孙进程继承导致 communicate 卡死, 故 stdin/stdout/stderr 全部与传输层隔离)。有头 Edge 指纹:知网(kns.cnki.net / bar.cnki.net)为瑞数级风控,无头浏览器直接 403; 本服务优先
channel="msedge"(真实 Edge 指纹)+ 有头运行,检索时桌面短暂弹窗属正常。v 令牌与会话绑定:检索结果的详情 url 带
v=令牌,跨浏览器会话会 403。 detail/download 均支持query(+title/index)参数:直开失败时在同一会话内重新检索并 点入结果行——即视频的原生流程。风控红线:安全验证=人工红线,截图停机(
captcha_required+ screenshot 路径), 绝不程序化破解。失败如实返回:
login_required/captcha_required/rate_limited/channel_error, 附 guidance 与截图路径。
环境变量
变量 | 默认 | 说明 |
|
|
|
| 项目 | 换新档案(被风控标记后换道重试) |
|
| kns8s 检索库 classid |
依赖
Python 3.12 + mcp(FastMCP)+ playwright(含 msedge channel)+ pymupdf。
This server cannot be deployed
Maintenance
Related MCP Connectors
Academic literature search, retrieval, and private library management on top of OpenAlex.
Federated search of books and papers, BibTeX/RIS citations, open-access retrieval and reading.
Search and download academic papers from arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, Semantic…
Search arXiv/Semantic Scholar/OpenAlex + medical evidence (PubMed/Europe PMC) + LaTeX/PDF tools.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to search CNKI (China National Knowledge Infrastructure), download PDFs, and import them into Zotero with automatic PDF attachment.32MIT
- FlicenseAqualityDmaintenanceEnables AI agents to search, read, and cite Chinese academic papers from CNKI using HUST single sign-on, with support for full-text retrieval, BibTeX export, and PDF download.74-
- AlicenseNot gradedqualityAmaintenanceEnables academic literature collection and full-text downloading from multiple sources (CNKI, Elsevier, OpenAlex, etc.) via natural language commands.3MIT
- AlicenseAqualityBmaintenanceEnables AI to search, filter, and process CNKI academic literature through natural language, including structured search, metadata retrieval, full-text reading, citation export, and PDF/CAJ download.98GPL 3.0