This MCP server exposes AgentQL's data-extraction capabilities as a reusable tool called extract-web-data. It enables clients to obtain structured data from unstructured web content by supplying a URL and a prompt that describes the data and fields to extract. The server is installable via npm (npm install -g agentql-mcp) and requires an AGENTQL_API_KEY for authentication, configured in the hosting environment or within IDE-specific settings. The README provides multiple integration paths across popular MCP-enabled environments: Claude, VS Code, Cursor, and Windsurf. Each platform includes concrete configuration blocks or steps to register the agentql MCP server under a named entry (e.g., agentql) with the appropriate command, arguments, and environment variables. Manual installation options include JSON blocks for Claude desktop settings, VS Code user settings (mcp.json), and a shared workspace file (.vscode/mcp.json). A sample usage demonstrates extracting a list of YouTube videos with title, author, views, and URL, formatted as a markdown table. Development and debugging guidance include building, watching for changes, and using the MCP Inspector for debugging over stdio.