MCP server that equips Claude with live web-research abilities (Google search, webpage scraping, screenshots, session tracking).
https://github.com/mzxrai/mcp-webresearchStop copying and pasting search results into Claude. This MCP server turns Claude into a research powerhouse with live Google search, webpage scraping, and screenshot capture—all without breaking your conversation flow.
You're deep in a Claude conversation about a technical implementation when you need current information—latest API changes, recent benchmarks, or breaking news in your domain. The flow breaks: copy question, open browser, search, scan results, copy relevant bits back to Claude, lose context, repeat.
The mcp-webresearch server eliminates this friction entirely. Claude gains direct access to current web information, maintains research context across multiple searches, and even captures visual evidence with screenshots.
Zero-Friction Setup: One line in your Claude Desktop config, and you're done. No API keys, no complex authentication—just npx and go.
Research Methodology Built-In: The included agentic-research prompt doesn't just search randomly. It guides Claude through systematic research: broad landscape searches first, authoritative source prioritization, iterative refinement based on findings.
Session Persistence: Unlike disconnected searches, this maintains a complete research trail—queries, visited pages, extracted content, screenshots, timestamps. Your research builds on itself.
Visual Context: Screenshots aren't just saved somewhere—they become MCP resources accessible directly in Claude Desktop. Perfect for UI research, design analysis, or when you need visual proof.
API Integration Research: "Research the latest changes to the Stripe API and show me any breaking changes in the past 6 months." Claude searches, visits the official changelog, captures relevant screenshots, and maintains the full research context.
Competitive Analysis: "Find recent performance benchmarks comparing Next.js 14 with Remix." Claude systematically searches authoritative sources, compiles findings, and provides screenshots of actual benchmark charts.
Breaking News Impact: "What's the latest on the Log4j vulnerability and current mitigation strategies?" Claude gets real-time information, visits security advisories, and maintains a complete research session you can reference later.
Documentation Deep-Dive: "Research how other companies implement rate limiting in their APIs and show me specific examples." Claude visits multiple implementation guides, captures code examples via screenshots, and builds a comprehensive comparison.
The beauty is in the seamlessness. Start any Claude conversation that needs current information, and Claude automatically uses web research when beneficial. No special commands, no switching contexts.
For deeper research sessions, access the built-in agentic-research prompt through Claude Desktop's paperclip menu. This transforms Claude into a methodical research assistant that keeps you informed at each step while building comprehensive knowledge systematically.
The session tracking means you can pause research, work on implementation, then return to the same Claude conversation and continue where you left off—complete with all previous searches, visited pages, and screenshots still accessible.
Add this single entry to your claude_desktop_config.json:
{
"mcpServers": {
"webresearch": {
"command": "npx",
"args": ["-y", "@mzxrai/mcp-webresearch@latest"]
}
}
}
Restart Claude Desktop. Start researching.
The server handles everything—Google searches, content extraction, HTML-to-markdown conversion, screenshot capture, and session management. Claude gains real-time web access without you managing any of the complexity.
Your research workflow just got a major upgrade. No more context switching, no more manual copy-paste research, no more losing track of what you've already investigated. Just Claude with real-time web intelligence, working the way research should work.