agentql-mcp
Overview
The agentql-mcp server is a Model Context Protocol (MCP) implementation that brings AgentQL’s web data extraction capabilities into your AI-driven workflows. It lets AI assistants retrieve structured data from web pages by accepting a URL and a natural-language description of the data you want. This means agents can incorporate real-time, prompt-driven web insights directly into automation, research, and content tasks without custom scraping scripts.king.
Transport
stdio
Tools
Key Capabilities
- Turn unstructured web pages into structured datasets by describing the target information in natural language.
- Support real-time data retrieval during agent workflows, so responses can include fresh, web-sourced content.
- Enable flexible prompt-driven extraction, letting you specify exactly what fields or elements to pull from a page.
- Integrate with MCP-compatible clients such as Claude, Cursor, Windsurf, and others to extend AI assistant capabilities.
- Open-source and configurable with a simple npm install and API key setup to tailor extraction workflows.
How It Works
The agentql-mcp server runs locally or within a configured environment and exposes AgentQL’s data extraction service to AI agents over the Model Context Protocol. Clients send MCP calls such as extract-web-data with a target URL and a prompt describing the desired fields. The server then reaches out to the specified web page, applies its AI-assisted extraction logic, and returns structured JSON results tailored to the prompt.
To use the server, you install it via npm, configure your environment with an AGENTQL_API_KEY, and add it to your MCP client’s settings (e.g., Claude Desktop or another MCP-aware application). Once started, the server listens for MCP requests and performs extraction without additional manual steps — turning arbitrary web URLs into machine-readable datasets inside your AI workflows.