Documentation | anybrowse

First scrape in 60 seconds

No signup. No API key. Pick your setup:

Add to your MCP config and ask your agent to scrape anything:

MCP config (all clients)

{
  "mcpServers": {
    "anybrowse": {
      "type": "streamable-http",
      "url": "https://anybrowse.dev/mcp"
    }
  }
}

Then ask your agent:

Natural language prompt

Scrape https://techcrunch.com and summarize the top stories

Where to put this config:

Client	Config location
Claude Code	`~/.claude/claude_desktop_config.json` → mcpServers
Cursor	Settings → MCP → Add Server
Windsurf	`~/.codeium/windsurf/mcp_config.json` → mcpServers
Claude Desktop	`~/Library/Application Support/Claude/claude_desktop_config.json`

curl

curl -X POST https://anybrowse.dev/scrape \
  -H "Content-Type: application/json" \
  -d '{"url": "https://techcrunch.com"}'

Response

# TechCrunch
## Top Stories

Latest news from the world of technology...
(returns full page as clean Markdown)

Python

import requests

r = requests.post("https://anybrowse.dev/scrape",
                  json={"url": "https://techcrunch.com"})
print(r.json()["markdown"])

Returns JSON for REST, plain text for MCP

REST API returns {"markdown": "..."}. MCP tools return plain markdown strings directly.

Try it live

WHAT HAPPENS NEXT

Free

10 scrapes/day. No API key. No signup. Resets at midnight UTC.

Pro

$4.99/mo

Unlimited scrapes. API key. Priority queue. Get started

Pay-per-use

$0.002

Per scrape via x402 micropayments. USDC on Base. No subscription. Set up

When you hit the free limit, you get a 402 with an upgrade link in the response. Rate limits reset daily at 00:00 UTC.

MCP SETUP

Claude Code

~/.claude/claude_desktop_config.json

{
  "mcpServers": {
    "anybrowse": {
      "type": "streamable-http",
      "url": "https://anybrowse.dev/mcp"
    }
  }
}

Or use a project-local .mcp.json file in your repo root with the same structure.

Cursor

Go to Settings → MCP → Add Server and paste:

Cursor MCP config

{
  "mcpServers": {
    "anybrowse": {
      "type": "streamable-http",
      "url": "https://anybrowse.dev/mcp"
    }
  }
}

Windsurf

Go to Settings → Model Context Protocol or edit directly:

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "anybrowse": {
      "type": "streamable-http",
      "url": "https://anybrowse.dev/mcp"
    }
  }
}

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "anybrowse": {
      "type": "streamable-http",
      "url": "https://anybrowse.dev/mcp"
    }
  }
}

No key needed

Free tier works out of the box. For Pro, add "apiKey": "ab_your_key" to the server config.

MCP TOOLS & API REFERENCE

POST/scrapeMCP tool: scrape

Convert any URL to clean, LLM-ready Markdown. 84% scrape success rate — works on JavaScript-heavy sites, Cloudflare-protected pages, and government sites. Uses real Chrome with stealth mode, automatic CAPTCHA solving, and proxy fallback.

Param	Type	Description
`url` required	string	URL to scrape (http/https)
`slow` optional	boolean	Wait extra time for JS-heavy sites (default: false)
`waitFor` optional	number	Custom JS wait in ms (default: 5000)

Example

curl -X POST https://anybrowse.dev/scrape \
  -H "Content-Type: application/json" \
  -d '{"url": "https://news.ycombinator.com"}'

# Returns: plain Markdown text of the page

POST/crawlMCP tool: crawl

Recursively crawl a site and return multiple pages as Markdown. Ideal for docs sites and product catalogs.

Param	Type	Description
`url` required	string	Starting URL
`maxPages` optional	number	Max pages to crawl (default: 10, max: 100)
`maxDepth` optional	number	Link depth from start URL (default: 2)
`sameDomain` optional	boolean	Only follow same-domain links (default: true)

Response (200 OK)

{
  "pages": [{ "url": "https://docs.example.com", "title": "Introduction", "markdown": "# Introduction..." }],
  "crawled": 5, "skipped": 1
}

POST/batchMCP tool: batch_scrape

Scrape up to 10 URLs in parallel in a single call. More efficient than looping /scrape.

Param	Type	Description
`urls` required	string[]	Array of URLs (max 10)

Example

curl -X POST https://anybrowse.dev/batch \
  -H "Content-Type: application/json" \
  -d '{"urls": ["https://techcrunch.com", "https://theverge.com", "https://wired.com"]}'

# Response: { "results": [{ "url": "...", "success": true, "markdown": "...", "title": "..." }] }

POST/extractMCP tool: extract

Extract structured JSON from any page using a field schema. Returns typed data alongside raw Markdown.

Param	Type	Description
`url` required	string	URL to extract from
`schema` required	object	Field names mapped to types: `string`, `number`, `boolean`, `array`

Example

curl -X POST https://anybrowse.dev/extract \
  -H "Content-Type: application/json" \
  -d '{"url": "https://shop.example.com/product/1",
       "schema": {"price": "number", "title": "string", "in_stock": "boolean"}}'

# Response: { "data": { "price": 29.99, "title": "Widget Pro", "in_stock": true }, "markdown": "..." }

POST/serp/searchMCP tool: search

Real-time web search via Brave Search API — fast, reliable, no rate limits. Returns results as structured JSON. Use before scraping to find the right pages.

Param	Type	Description
`query` required	string	Search query
`num` optional	number	Number of results (default: 10)

Example

curl -X POST https://anybrowse.dev/serp/search \
  -H "Content-Type: application/json" \
  -d '{"query": "best MCP servers 2025", "num": 5}'

# Response: { "results": [{ "title": "...", "url": "...", "snippet": "..." }] }

ERROR REFERENCE

Status	Error	Fix
402	Daily limit reached	Upgrade to Pro at /checkout or use x402 pay-per-use
402	Payment verification failed	Check wallet balance; use the x402-fetch library (see below)
200	Empty or minimal content	Site may require more JS time. Add `"slow": true` to request
451	BLOCKED	Site blocked scraping after proxy fallback. Try a different URL or contact us
429	TOO_MANY_REQUESTS	Check the `Retry-After` header and wait before retrying
504	TIMEOUT	Page took over 30s to load. Try with a shorter `timeout` or skip
500	INTERNAL_ERROR	Retry once; if persistent the URL may be permanently blocked
400	BAD_REQUEST	Check that `url` starts with `http://` or `https://`

Error response shape

{ "error": "Human-readable message", "reason": "REASON_CODE", "upgrade": "https://anybrowse.dev/checkout" }

x402 MICROPAYMENTS

Pay $0.002/scrape in USDC on Base. No subscription, no account needed. The x402-fetch library handles payment automatically.

Install

npm install x402-fetch

JavaScript / TypeScript

import { x402Fetch } from 'x402-fetch';

const client = x402Fetch({
  wallet: { privateKey: process.env.WALLET_PRIVATE_KEY, network: 'base' },
  facilitator: 'coinbase'
});

const res = await client('https://anybrowse.dev/scrape', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({ url: 'https://techcrunch.com' })
});
const markdown = await res.text();

Test for free on Base Sepolia

Use network: 'base-sepolia' and get test USDC from the Circle faucet.

RATE LIMITS

Endpoint	Free	Pro	x402
`/scrape`	10/day	10,000/month	Unlimited (pay per call)
`/crawl`	10/day	10,000/month	Unlimited
`/batch`	5/day	Unlimited	Unlimited
`/serp/search`	10/day	10,000/month	Unlimited
`/mcp` (connect)	Free	Free	Free

Limits reset at midnight UTC. Rate limit responses include a Retry-After header in seconds.