Model Context Protocol server that integrates AgentQL's data-extraction capabilities and exposes the `extract-web-data` tool.
https://github.com/tinyfish-io/agentql-mcpStop wrestling with brittle CSS selectors and DOM mutations. This MCP server brings AgentQL's AI-powered data extraction directly into your Claude, VS Code, or Cursor workflow—just describe what you want in plain English.
You know the drill: spend 30 minutes crafting the perfect CSS selector, deploy your scraper, then watch it break when the site updates their class names next week. Traditional scraping tools force you to think like a DOM parser instead of focusing on the actual data you need.
AgentQL flips this on its head. Instead of div.video-card > h3.title-text > a[href*="/watch"], you just say "get the video title and URL." The AI figures out the selectors, handles dynamic content, and adapts when sites change their structure.
One Tool, Maximum Impact: The extract-web-data tool takes a URL and a natural language description of what you want. That's it.
Extract the list of videos from YouTube search results.
Each video should have title, author, view count, and URL.
Exclude sponsored content.
No XPath. No CSS inspection. No "wait, why is this div suddenly empty?" debugging sessions.
Real Browser Execution: This isn't parsing static HTML—it uses AgentQL's browser automation to handle JavaScript rendering, lazy loading, and dynamic content. If a human can see it, AgentQL can extract it.
Production-Ready Integration: Works seamlessly with Claude Desktop, VS Code, Cursor, and Windsurf. One npm install, add your API key, and you're extracting data in your next conversation.
E-commerce Intelligence: "Get product names, prices, ratings, and availability from this Amazon search page" returns clean structured data without reverse-engineering their anti-bot measures.
Content Research: Pull article headlines, author info, and publication dates from any news site. Works across different CMSs without site-specific adaptations.
Lead Generation: Extract company listings, contact information, and business details from directory sites that would normally require manual copy-paste work.
Competitive Analysis: Monitor pricing pages, feature lists, and product updates across competitor sites with simple English descriptions instead of maintenance-heavy scraping scripts.
Social Media Monitoring: Pull post content, engagement metrics, and profile information from platforms where traditional APIs are restricted or expensive.
npm install -g agentql-mcp
Add to your Claude config:
{
"mcpServers": {
"agentql": {
"command": "npx",
"args": ["-y", "agentql-mcp"],
"env": {
"AGENTQL_API_KEY": "YOUR_API_KEY"
}
}
}
}
Start extracting data:
Extract pricing tiers from https://openai.com/pricing -
include plan name, price, and key features for each tier
This isn't just another scraping tool—it's a fundamental shift in how you interact with web data. Instead of spending hours debugging selectors and handling edge cases, you describe what you want and get structured JSON back. Your AI assistant can now gather market research, compile competitive intelligence, or build datasets with the same ease it analyzes your local files.
Time to extraction: under 5 minutes. Time saved per scraping task: hours.
Get your AgentQL API key: dev.agentql.com