A Model-Context-Protocol server that performs parallel Google searches via Playwright and returns structured JSON results.
https://github.com/jae-jae/g-search-mcpYour AI assistant just got a search upgrade. While other tools crawl through Google one query at a time, G-Search MCP runs multiple searches simultaneously and delivers structured results in seconds.
Most developers resort to Google's API (with usage limits and costs) or basic scraping tools that process searches sequentially. When you're researching multiple topics or comparing different angles on a problem, you're stuck waiting for each search to complete before the next one starts.
That's painfully slow when you need to gather information on 5-10 related topics for a comprehensive analysis.
G-Search MCP opens multiple browser tabs in a single Chromium instance and runs your searches concurrently. Ask Claude to research "React performance optimization," "Vue 3 composition API," and "Svelte compiler advantages" - and get all three result sets back in the time it normally takes to complete one search.
The tool simulates real browsing behavior to avoid detection, automatically handles CAPTCHAs when they appear, and returns clean JSON with titles, URLs, and snippets ready for your AI to analyze.
Here's how it fits into your actual development process:
Competitive Analysis: "Search for 'headless CMS comparison 2024', 'Strapi vs Sanity', and 'GraphQL CMS solutions' - give me the top 10 results for each"
Technical Research: "Find recent discussions about 'Next.js app router performance', 'React Server Components adoption', and 'Vercel edge functions limitations'"
Problem Solving: "Search for 'TypeScript generic constraints', 'conditional types examples', and 'mapped types utility' - I need comprehensive examples for each"
Each request returns structured data that Claude can immediately analyze, summarize, and synthesize into actionable insights.
Add this to your Claude Desktop config and you're done:
{
"mcpServers": {
"g-search": {
"command": "npx",
"args": ["-y", "g-search-mcp"]
}
}
}
No API keys, no registration, no usage limits. The first run installs Playwright's Chromium browser automatically.
The tool includes practical configuration options you'll actually use:
Want to see what's happening? Add --debug to watch the searches execute in real-time.
Unlike generic scraping tools, this is designed specifically for the research patterns developers use. It understands that you often need to:
The parallel execution means your research sessions are constrained by your thinking time, not by sequential API calls.
npx playwright install chromium
Then configure Claude Desktop and start asking for multi-topic research. Your first parallel search will demonstrate why sequential tools feel broken once you've experienced this approach.
Repository: github.com/jae-jae/g-search-mcp
The 160+ GitHub stars suggest other developers have already discovered how much faster their research workflows become with parallel search capability built directly into their AI assistant.