CrawlAPI MCP Server

crawlapi-js

Official JavaScript/TypeScript SDK for CrawlAPI — web scraping API built for AI agents.

Pass a URL, get back clean markdown your LLM can actually read.

Install

npm install crawlapi-js

Quick Start

const CrawlAPI = require('crawlapi-js');

const client = new CrawlAPI({ apiKey: 'YOUR_RAPIDAPI_KEY' });

// Scrape a URL → clean markdown
const result = await client.scrape('https://example.com');
console.log(result.data.markdown);

// Scrape multiple URLs in parallel
const batch = await client.batch([
  'https://example.com',
  'https://example.org'
]);

// Search + scrape top results in one call
const search = await client.search('LangChain web scraping 2025', { num: 5 });

Get an API Key

Available on RapidAPI. Free tier: 50 calls/day, no credit card.

API

`client.scrape(url, options?)`

Scrape a single URL with full JavaScript rendering.

const result = await client.scrape('https://example.com', {
  formats: ['markdown'],   // 'markdown' | 'html' | 'text' | 'structured'
  waitFor: 1000,           // ms to wait after page load (JS-heavy pages)
  timeout: 30000           // max ms (default 30s, max 60s)
});

console.log(result.data.markdown);
console.log(result.data.metadata.title);

`client.batch(urls, options?)`

Scrape up to 10 URLs in parallel. Failed URLs return an error field rather than failing the whole request.

const result = await client.batch([
  'https://example.com',
  'https://example.org',
  'https://example.net'
]);

result.data.forEach(item => {
  if (item.success) console.log(item.data.markdown);
  else console.error(item.url, item.error);
});

`client.search(query, options?)`

Search the web and automatically scrape the top N results.

const result = await client.search('best pizza in New York', {
  num: 5,                  // 1-10 results (default 5)
  formats: ['markdown']
});

result.data.forEach(item => {
  console.log(item.title, item.url);
  if (item.success) console.log(item.data.markdown);
});

Agent Framework Examples

LangChain

const { Tool } = require('langchain/tools');
const CrawlAPI = require('crawlapi-js');

const client = new CrawlAPI({ apiKey: process.env.RAPIDAPI_KEY });

const webTool = new Tool({
  name: 'WebScraper',
  description: 'Scrape any URL and return clean markdown content. Input should be a URL.',
  func: async (url) => {
    const result = await client.scrape(url);
    return result.data.markdown;
  }
});

OpenAI Function Calling

const CrawlAPI = require('crawlapi-js');
const client = new CrawlAPI({ apiKey: process.env.RAPIDAPI_KEY });

// Tool definition
const tools = [{
  type: 'function',
  function: {
    name: 'scrape_url',
    description: 'Scrape a webpage and return its content as clean markdown',
    parameters: {
      type: 'object',
      properties: {
        url: { type: 'string', description: 'The URL to scrape' }
      },
      required: ['url']
    }
  }
}];

// Handle tool call
async function handleToolCall(url) {
  const result = await client.scrape(url);
  return result.data.markdown;
}

n8n / Flowise / Custom HTTP

All endpoints accept standard JSON POST requests:

POST https://crawlapi.net/v1/scrape
X-RapidAPI-Key: YOUR_KEY
Content-Type: application/json

{ "url": "https://example.com", "formats": ["markdown"] }

Full OpenAPI spec: crawlapi.net/openapi.json

TypeScript

Full TypeScript types included:

import { CrawlAPI, ScrapeResponse } from 'crawlapi-js';

const client = new CrawlAPI({ apiKey: process.env.RAPIDAPI_KEY! });
const result: ScrapeResponse = await client.scrape('https://example.com');

MCP Server (Claude Desktop, Cursor, Windsurf)

CrawlAPI ships with a built-in MCP server — plug it directly into any MCP-compatible AI client.

Setup

1. Clone or download mcp-server.js from this repo.

2. Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on Mac):

{
  "mcpServers": {
    "crawlapi": {
      "command": "node",
      "args": ["/path/to/crawlapi-js/mcp-server.js"],
      "env": {
        "CRAWLAPI_KEY": "your_rapidapi_key"
      }
    }
  }
}

3. Restart Claude Desktop. You'll see three new tools:

scrape_url — scrape any page to markdown
batch_scrape — scrape up to 10 URLs in parallel
search_and_scrape — search + scrape top results in one shot

Cursor / Windsurf / Continue

Same config, different location. Check your editor's MCP docs. The mcp-server.js file works with any stdio-transport MCP client.

What the agent sees

"Use scrape_url to fetch web pages as clean markdown. Use search_and_scrape to research topics from the live web. Both support JavaScript-rendered pages."

Get your API key at RapidAPI — free tier included.

AI Agent Discovery

CrawlAPI publishes all standard discovery files:

File	URL
`llms.txt`	crawlapi.net/llms.txt
OpenAPI spec	crawlapi.net/openapi.json
AI Plugin manifest	crawlapi.net/.well-known/ai-plugin.json
MCP manifest	crawlapi.net/.well-known/mcp.json
APIs.json	crawlapi.net/apis.json

Documentation