Prompt & Done · SCRAPEBEE setup
One sentence to Claude Code, and the Hacker News front page became a CSV.
Scrapling is a free, open-source (BSD-3-Clause) web-scraping library with an MCP server, so Claude Code can fetch and extract pages for you. Tested 26 Oct 2026 with Scrapling 0.5: one prompt, 30 rows, 6 columns, 21 seconds, all 30 titles, authors and points matching the live page.
1 Install it in its own folder (nothing global)
python3 -m venv .venv && source .venv/bin/activate
pip install "scrapling[ai]"Source and docs: github.com/D4Vinci/Scrapling. The browser-based fetchers need an extra scrapling install download; the demo here used the plain HTTP tool and did not need it.
2 Give Claude Code the MCP server (this project only)
Create .mcp.json in your project folder (the file in the download, with your venv path):
{"mcpServers":{"scrapling":{"command":"/full/path/to/.venv/bin/scrapling","args":["mcp"]}}}Then start claude in that folder and approve the project MCP server when asked.
3 The prompt, exactly as in the reel
Use Scrapling to scrape the Hacker News front page into a CSV with rank, title, site, points, comments and author.Name Scrapling in the prompt: when we left it out, Claude Code used curl and a script instead. The CSV from our run is in the download (hn_frontpage.csv).
4 Honest limits
One run on one site. Hacker News' robots.txt allows the front page (it asks for a 30-second crawl delay, so fetch it once, not in a loop). Always check a site's robots.txt and terms before scraping, and skip logins and personal data. We did not test anti-bot or Cloudflare-protected pages, so we make no claim about them. Claude typed the values into the CSV itself rather than running a script, so spot-check important data.