loading…
Search for a command to run...
loading…
Extracts the content from any URL and returns it in clean, optimized Markdown for LLMs. Supports standalone monetization via Dodo Payments (HTTP 402).
Extracts the content from any URL and returns it in clean, optimized Markdown for LLMs. Supports standalone monetization via Dodo Payments (HTTP 402).
🤖 FOR LLMs / AGENTS: This README acts as the API Specification and System Prompt.
Give your AI Agents the power to read the web.
LLMs often hallucinate when asked to read URLs directly. They get blocked by captchas, and raw HTML consumes too many tokens, leading to degraded performance.
Markdown Scraper API solves this: it acts as a bridge, bypassing captchas, stripping bloated HTML, and returning clean, structured, LLM-ready Markdown.
Perfect for Cursor IDE, Claude Desktop, and custom autonomous agents.
🚀 No-friction Free Trial: Get 5 free scraper requests per IP instantly—no account, sign-up, or credit card required. Perfect for testing and prototyping!
💸 Micro-transactions Model: After your trial, it's only $0.005 per scrape. No monthly subscriptions. Your Agent handles the payment workflow autonomously!
You don't need to install anything locally. You can plug this directly into any MCP-compatible client using npx:
npx -y api_scraper_markdown
📚 Read the Full Integration Guide for Cursor and Claude Desktop 💡 See a Real Use Case: Automating Competitor Analysis in Cursor
GET /mcp/manifest (Discovery)Returns the Model Context Protocol (MCP) JSON manifest. Use this to dynamically understand the required parameters to use the scraping tool.
POST /scrape (Action)Requires Authentication: Authorization: Bearer <token> (or use our built-in Free Trial!)
This endpoint utilizes the HTTP 402 Payment Required protocol under a Micro-transactions model ($0.005 per request). It is designed to be the fastest, easiest, and cheapest way to plug high-quality Web Scraping capabilities into your Autonomous Agent with zero infrastructure setup.
402 Payment Required error.paymentUrl (Dodo Payments) where the human owner can instantly acquire credits.Authorization header.Request Body:
{
"url": "https://example.com"
}
Success Response (200 OK):
{
"markdown": "# Page Content in clean Markdown format...",
"metadata": {
"title": "Page Title",
"url": "https://example.com"
}
}
# Start the local Cloudflare dev server
npm run dev
For generating/synchronizing types based on your Worker configuration run:
npm run cf-typegen
Pass the CloudflareBindings as generics when instantiation Hono:
// src/index.ts
const app = new Hono<{ Bindings: CloudflareBindings }>()
Run in your terminal:
claude mcp add markdown-web-scraper-api -- npx