scrapi-mcp and crawl4ai-mcp-server
These are competitors offering alternative approaches to the same problem—both wrap different web scraping engines (ScrAPI vs. Crawl4AI) as MCP servers to expose scraping capabilities to AI agents, so users would typically choose one based on their preferred underlying scraper's performance and feature set.
About scrapi-mcp
DevEnterpriseSoftware/scrapi-mcp
MCP server for using ScrAPI to scrape web pages.
Exposes two scraping tools that return content in HTML or Markdown formats, with support for browser automation commands (clicks, form filling, scrolling, JavaScript execution) to handle dynamic content and bot detection. Integrates with Claude Desktop and other MCP clients via stdio transport, with optional cloud deployment over SSE, and requires a ScrAPI API key for production use beyond free tier limitations.
About crawl4ai-mcp-server
sadiuysal/crawl4ai-mcp-server
🕷️ A lightweight Model Context Protocol (MCP) server that exposes Crawl4AI web scraping and crawling capabilities as tools for AI agents. Similar to Firecrawl's API but self-hosted and free. Perfect for integrating web scraping into your AI workflows with OpenAI Agents SDK, Cursor, Claude Code, and other MCP-compatible tools.
Implements a stdio-based MCP server using Playwright for browser automation, exposing granular tools for single-page scraping (`scrape`), multi-page breadth-first crawling (`crawl`), and large-scale sitemaps (`crawl_site`/`crawl_sitemap`) with optional persistent storage to disk. Supports adaptive crawling that intelligently stops when sufficient content is gathered, safety constraints blocking private IPs, and integration with OpenAI Agents SDK, Cursor, and Claude Code through standardized MCP configuration.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work