AgentQL MCP Server: Structured Web Data for Claude, Cursor, Windsurf, and more

AgentQL now supports Model Context Protocol (MCP), making it easier than ever to integrate live, structured web data into tools that support MCP—like Claude, Cursor, and Windsurf. MCP is an open standard that allows AI assistants to use external tools. With the AgentQL MCP Server, your AI models can now extract live data from web pages—transforming unstructured HTML into structured, usable data. Every web page: an endpoint Before today, most AI models were limited to static knowledge or aging web scrapes, and much content on the web is provided without an API to access. Now, with AgentQL + MCP, your assistant can: Extract data from unstructured HTML websites Extract just the content from websites without ads or UI Pull live YouTube search results, filtering out ads Business Intelligence from recent news and markets Get pricing data from different marketplaces Collect financial report information from public web pages Research with fresh data from the web Include recent news from outlet sites in the context of your conversation Retrieve data from online articles All in real-time. No scraping scripts. No extra APIs. No vector storage. Set Up AgentQL MCP Server To get started, install the AgentQL MCP package: npm install -g agentql-mcp Configure Claude with AgentQL MCP Server Configure Cursor with AgentQL MCP Server Configure Windsurf with AgentQL MCP Server Try out AgentQL MCP Server today Try running some of the following commands: Get the social links from agentql.com Return the top five video titles and their links from https://www.youtube.com/results?search_query=agentql in a list in markdown Return the top international news articles from ground.news into a CSV And, in our opinion AgentQL’s “killer usecase”—using Claude as a cookbook: Get this recipe for me: https://www.justonecookbook.com/pressure-cooker-japanese-curry/ Now your AI models can retrieve structured web data in real-time—without needing a prebuilt API or manual scraping. Open to collaborations The AgentQL MCP Server is opensource, and we value the community’s feedback and contributions. We work hard to make tools that let your agents reach beyond their foundation model’s knowledge and leverage the whole world wide web. If you’ve got ideas or challenges, we want to hear them and solve them with you. We can't wait to see what you build! Give us a shout out on Discord, X, or Bluesky! —The TinyFish team building AgentQL

Mar 13, 2025 - 05:59

AgentQL MCP Server: Structured Web Data for Claude, Cursor, Windsurf, and more

AgentQL now supports Model Context Protocol (MCP), making it easier than ever to integrate live, structured web data into tools that support MCP—like Claude, Cursor, and Windsurf.

MCP is an open standard that allows AI assistants to use external tools. With the AgentQL MCP Server, your AI models can now extract live data from web pages—transforming unstructured HTML into structured, usable data.

Every web page: an endpoint

Before today, most AI models were limited to static knowledge or aging web scrapes, and much content on the web is provided without an API to access. Now, with AgentQL + MCP, your assistant can:

Extract data from unstructured HTML websites
- Extract just the content from websites without ads or UI
- Pull live YouTube search results, filtering out ads
Business Intelligence from recent news and markets
- Get pricing data from different marketplaces
- Collect financial report information from public web pages
Research with fresh data from the web
- Include recent news from outlet sites in the context of your conversation
- Retrieve data from online articles

All in real-time. No scraping scripts. No extra APIs. No vector storage.

Set Up AgentQL MCP Server

To get started, install the AgentQL MCP package:

npm install -g agentql-mcp

Try out AgentQL MCP Server today

Try running some of the following commands:

Get the social links from agentql.com
Return the top five video titles and their links from https://www.youtube.com/results?search_query=agentql in a list in markdown
Return the top international news articles from ground.news into a CSV

And, in our opinion AgentQL’s “killer usecase”—using Claude as a cookbook:

Get this recipe for me: https://www.justonecookbook.com/pressure-cooker-japanese-curry/

Now your AI models can retrieve structured web data in real-time—without needing a prebuilt API or manual scraping.

Open to collaborations

The AgentQL MCP Server is opensource, and we value the community’s feedback and contributions. We work hard to make tools that let your agents reach beyond their foundation model’s knowledge and leverage the whole world wide web. If you’ve got ideas or challenges, we want to hear them and solve them with you.

We can't wait to see what you build! Give us a shout out on Discord, X, or Bluesky!

—The TinyFish team building AgentQL