SCI Search Skill

SCI-search-skill

A [Claude Code](https://claude.com/claude-code) skill that searches academic databases and downloads SCI papers through **your own everyday Chrome browser** — not a headless bot.

Because websites see a real user, your login sessions, cookies, and institutional access all keep working. When verification or CAPTCHA appears, you complete it manually in the browser window while the script auto-detects completion and resumes.

Features

• 🔍 Search academic databases — Web of Science, Wiley Online Library, IEEE, Google Scholar, and more
• 🍪 Reuse your real sessions — logged-in cookies, institutional proxy access, and saved passwords all work
• 🤖 Captcha / Cloudflare handling — detects "verify you are human" / Cloudflare challenges and auto-resumes after you solve them
• 📥 Multi-strategy PDF download — browser-triggered download, requests with browser cookies, and XHR base64 for S3 signed URLs
• 📚 Web of Science specifics — fielded search, virtual-scrolling pagination, and metadata parsing from unstructured record text
• 🔁 State persistence — resume long download sessions after interruption

How it works

The skill drives your real Chrome instance over the Chrome DevTools Protocol (CDP), using a remote debugging port:

┌─────────────────────────────────────────────────────────┐
│  Your Chrome (--remote-debugging-port=9222)              │
│  ┌───────────────────────────────────────────────────┐  │
│  │  Tabs with your cookies, logins, sessions         │  │
│  └───────────────────────────────────────────────────┘  │
│                         ▲                                │
│                         │ CDP (WebSocket)                │
│                         ▼                                │
│  Python helper (references/cdp.py)                       │
│  - new_tab(url) → open & navigate                       │
│  - tab.eval(js) → execute JS, get results               │
│  - tab.navigate(url) → navigate existi