webclaw

MCP server

Give Claude real-time web access

One MCP server, and Claude can read any webpage — in the app, in the terminal, or in your editor.

Claude speaks the Model Context Protocol natively, so webclaw plugs straight in and Claude gets 12 web tools immediately: scrape, search, crawl, map, batch, extract, summarize, diff, brand, research, list_extractors, and vertical_scrape. The same server backs the desktop app, Claude Code and every editor extension — configure it once. No middleware, no custom code.

You are set up. Now try it.

Check it connected

Run it inside Claude Code and webclaw should be listed with 12 tools. In the desktop app, restart it and the tools appear in the tool menu.

Then paste this

PROMPT FOR YOUR AGENT

Use webclaw to scrape https://news.ycombinator.com, then give me the five posts most relevant to AI infrastructure with a one-line summary of each and its link.

Why webclaw for Claude

  • Native MCP server, no wrappers
  • 12 tools exposed in one config
  • Handles bot-protected sites automatically
  • Clean markdown output fits agent context windows

Common use cases

  • Read current framework docs while editing code
  • Research topics across the web without copy-pasting tabs
  • Scrape API references, GitHub issues, changelogs and release notes
  • Extract structured data from public pages into code, tests or notes

FAQ

Frequently asked questions.

Book a call

Run claude mcp add webclaw -- npx -y @webclaw/mcp. Pass WEBCLAW_API_KEY as an environment variable, and use --scope user so it is available across projects. For the desktop app, add the JSON block above to claude_desktop_config.json and restart it.

Ready to connect?

Start extracting.

Claude website