wos-skills

Web of Science Skills for Claude Code

English | 中文

WeChat Official Account (公众号)	WeChat Group (微信群)	Discord
		Join Discord
未来论文实验室	扫码加入交流群	English & Chinese

---

English

Claude Code skills that let Claude interact with Web of Science (WoS) through Chrome DevTools MCP.

Search papers, extract metadata, navigate results, export citations, download PDFs, and push to Zotero — all from the Claude Code CLI.

Prerequisites

• Claude Code CLI installed
• Chrome browser (institutional login for WoS access)
• Zotero desktop app (optional, for citation export)
• Python 3 (optional, for Zotero push script)

Skills

Skill	Description	Invocation
wos-search	Search WoS by topic, author, title, DOI. Supports edition/database filtering and sort	/wos-search value co-creation --edition SSCI --sort citations
wos-paper-detail	Extract full record metadata (abstract, JIF, keywords, citations, etc.)	/wos-paper-detail WOS:000295471900004
wos-navigate-pages	Paginate through search results via API or URL	/wos-navigate-pages 2
wos-parse-results	Parse search results from API response or DOM	(internal)
wos-download	Download article PDFs via publisher links	/wos-download WOS:000295471900004
wos-export	Export citations to Zotero, RIS, BibTeX, or Excel	/wos-export zotero

Agent

wos-researcher — orchestrates all skills. Supports multi-step workflows like "search → detail → export to Zotero → download PDF". API-first approach with DOM fallback.

Key Features

• API-first: Uses WoS internal API (runQuerySearch) for search and pagination — 1 tool call, structured JSON, no DOM fragility
• Edition filtering: Search within SCI, SSCI, CPCI, or any combination
• Sort control: By citations, date, relevance, or usage
• Multiple databases: Core Collection (default), All Databases, MEDLINE, Preprint, SciELO, etc.
• JIF & JCR extraction: Impact Factor, Five-Year IF, and JCR Quartile from full record pages
• Zotero integration: Direct push via Connector API with deterministic session IDs (no duplicates)
• Chinese keyword translation: Auto-translates Chinese search terms to English for SCI/SSCI

Installation

1. Install Chrome DevTools MCP server

claude mcp add chrome-devtools -- npx -y chrome-devtools-mcp@latest

2. Install WoS skills

git clone https://github.com/cookjohn/wos-skills.git
cd wos-skills
cp -r skills/ agents/ .claude/

Or add to an existing project:

git clone https://github.com/cookjohn/wos-skills.git /tmp/wos-skills
cp -r /tmp/wos-skills/skills/ your-project/.claude/skills/
cp -r /tmp/wos-skills/agents/ your-project/.claude/agents/

3. Configure anti-detection (recommended)

Add to your .claude.json to prevent bot detection:

{
  "mcpServers": {
    "chrome-devtools": {
      "command": "npx",
      "args": [
        "-y",
        "chrome-devtools-mcp@latest",
        "--ignoreDefaultChromeArg=--enable-automation",
        "--ignoreDefaultChromeArg=--disable-infobars",
        "--chromeArg=--disable-blink-features=AutomationControlled"
      ]
    }
  }
}

4. Log in to WoS

Before using the skills, open Chrome and log in to Web of Science with your institutional account. The skills require an active WoS session.

Usage Examples

# Simple topic search
/wos-search deep learning

# Search SSCI for value co-creation, sorted by citations
/wos-search value co-creation --edition SSCI --sort citations

# Multi-field advanced search
/wos-search (builds: TS="AI" AND AU="Smith" AND PY=2020-2025)

# View paper details with JIF and JCR quartile
/wos-paper-detail WOS:000295471900004

# Export to Zotero
/wos-export zotero

# Navigate to page 2
/wos-navigate-pages 2

Architecture

wos-skills/
├── skills/
│   ├── wos-search/SKILL.md          # API-based search (1 tool call)
│   ├── wos-paper-detail/SKILL.md    # Full record extraction (2 tool calls)
│   ├── wos-navigate-pages/SKILL.md  # API pagination (1 tool call)
│   ├── wos-parse-results/SKILL.md   # Internal result parser
│   ├── wos-download/SKILL.md        # PDF download via publisher
│   ├── wos-export/
│   │   ├── SKILL.md                 # Export & Zotero push
│   │   └── scripts/
│   │       └── push_to_zotero.py    # Zotero Connector API script
│   └── wos-dom.md                   # DOM & API reference document
└── agents/
    └── wos-researcher.md            # Orchestrator agent

Related Projects

• sd-skills — ScienceDirect skills for Claude Code

---

中文

Claude Code 技能，通过 Chrome DevTools MCP 让 Claude 与 Web of Science (WoS) 交互。

搜索论文、提取元数据、浏览结果、导出引用、下载 PDF、推送到 Zotero —— 全部通过 Claude Code CLI 完成。

前提条件

• 已安装 Claude Code CLI
• Chrome 浏览器（需要机构登录访问 WoS）
• Zotero 桌面版（可选，用于引用导出）
• Python 3（可选，用于 Zotero 推送脚本）

技能列表

技能	描述	调用方式
wos-search	按主题、作者、标题、DOI 搜索，支持版本/数据库筛选和排序	/wos-search 价值共创 --edition SSCI --sort citations
wos-paper-detail	提取论文详细信息（摘要、JIF、关键词、引用数等）	/wos-paper-detail WOS:000295471900004
wos-navigate-pages	通过 API 或 URL 翻页	/wos-navigate-pages 2
wos-parse-results	解析搜索结果（内部技能）	（内部使用）
wos-download	通过出版商链接下载 PDF	/wos-download WOS:000295471900004
wos-export	导出引用到 Zotero、RIS、BibTeX 或 Excel	/wos-export zotero

核心特性

• API 优先：使用 WoS 内部 API 搜索和翻页 —— 1 次工具调用，结构化 JSON，无 DOM 脆弱性
• 版本筛选：在 SCI、SSCI、CPCI 或任意组合中搜索
• 排序控制：按引用数、日期、相关性或使用量排序
• 多数据库：核心合集（默认）、所有数据库、MEDLINE、预印本、SciELO 等
• 影响因子提取：从全记录页面提取 JIF、五年 IF 和 JCR 分区
• Zotero 集成：通过 Connector API 直接推送，确定性会话 ID 防止重复
• 中文关键词翻译：自动将中文搜索词翻译为英文用于 SCI/SSCI 搜索

安装

1. 安装 Chrome DevTools MCP 服务器

claude mcp add chrome-devtools -- npx -y chrome-devtools-mcp@latest

2. 安装 WoS 技能

git clone https://github.com/cookjohn/wos-skills.git
cd wos-skills
cp -r skills/ agents/ .claude/

3. 配置反检测（推荐）

在 .claude.json 中添加以下配置：

{
  "mcpServers": {
    "chrome-devtools": {
      "command": "npx",
      "args": [
        "-y",
        "chrome-devtools-mcp@latest",
        "--ignoreDefaultChromeArg=--enable-automation",
        "--ignoreDefaultChromeArg=--disable-infobars",
        "--chromeArg=--disable-blink-features=AutomationControlled"
      ]
    }
  }
}

4. 登录 WoS

使用前请先在 Chrome 中用机构账号登录 Web of Science。

相关项目

• sd-skills — ScienceDirect 学术网站 Claude Code 技能