Crawl4ai MCP Server

Example Use Cases

• 1

  Fetch and aggregate web page content into markdown with clear per-URL citations for downstream LLM processing.

• 2

  Crawl multiple URLs programmatically and receive consolidated, LLM-friendly formatted results for summarization or question answering.

• 3

  Integrate crawl4ai's crawling capabilities as an MCP endpoint within an agent orchestration system or local toolchain.

Description

This repository provides a Node.js MCP server that connects to a crawl4ai instance to crawl URLs and emit markdown output with citation footnotes, suitable for LLM consumption. It exposes a tool called crawl_urls which accepts a list of URLs and returns markdown-separated results with source citations. Configuration is done via environment variables (CRAWL4AI_API_URL and optional CRAWL4AI_AUTH_TOKEN) and the project is distributed under the MIT License. It is intended to be run alongside a crawl4ai service (documentation link included) and can be used within agent frameworks that support MCP servers.