Scoop 🐼

Extract your LinkedIn saved posts into a structured research index.

What it does

• Connects to Chrome via DevTools Protocol (CDP)

• Clicks "Show more results" to load all saved posts

• Extracts author names, timestamps, post text, external links

• Detects "link in comments" patterns for GitHub repos

• Outputs structured JSON and generates a Markdown index

Related MCP server: LinkedIn MCP Server

Setup

1. Create a dedicated Chrome profile

& "C:\Program Files\Google\Chrome\Application\chrome.exe" `
  --remote-debugging-port=5192 `
  --remote-debugging-address=127.0.0.1 `
  --user-data-dir="$env:LOCALAPPDATA\Google\Chrome\Slot4" `
  --profile-directory=Default `
  --no-first-run `
  --no-default-browser-check `
  "https://www.linkedin.com/login"

Sign into LinkedIn in the Chrome window that opens. The session persists in Slot4.

2. Add the MCP server

Add to your .mcp.json or Zed's settings.json:

"chrome-research": {
  "command": "npx",
  "args": [
    "-y", "chrome-devtools-mcp@latest",
    "--browserUrl", "http://127.0.0.1:5192",
    "--no-usage-statistics",
    "--experimental-page-id-routing"
  ]
}

3. Sync your saved posts

Launch Chrome with the research profile, then tell your AI agent:

"Sync my saved articles"

The agent will use the chrome_research_* MCP tools to scrape all posts and generate your research index.

Architecture

Chrome (Slot4) → DevTools port 5192 → MCP server → AI agent → Research index

Files

• scripts/launch.ps1 — Launch Chrome with the research profile

• scripts/scrape.mjs — CDP-based LinkedIn saved posts scraper

• lib/cdp-client.mjs — Reusable WebSocket/CDP client

• lib/extractor.mjs — Post extraction logic

License

MIT