100 free credits · no card required

Turn the live web into agent-ready data.

Scrape, crawl, search and extract any site. One API, built for agents.

Start for free Connect your agent

Typed SDKs in 9 languages,

plus a remote MCP server for agents.

[ 01 / 07 ]Capabilities

Everything between a URL and usable data.

One key, one credit balance, one response shape.

Scrape

1 credit

Turn one page into clean markdown, HTML, links or JSON.

HTML → MARKDOWN

Crawl & map

1 credit / page

Discover and read every page across a site as one job.

1,284 pages discovered

Search

2 credits

Search the live web and get structured sources back.

QUERY → SOURCES

Extract

1 credit / page

Describe the data you need and get typed JSON back.

SCHEMA → JSON

Screenshot

5 credits

Capture a clean page image with banners and overlays removed.

1440 × 900 · PNG

Brand

10 credits

Get a company's logos, colors, fonts and socials from its domain.

FONT · COLOR · LOGO

Also: YouTube transcripts · PDF, DOCX and XLSX parsing · async jobs · signed webhooks

[ 02 / 07 ]Reliability

Every page rendered, cleaned and checked.

Rendering, proxy escalation and cleanup run on every request.

Failed requests

0credits

for blocked, empty or error pages. You get an error code, never junk content.

Blocked
Empty
Error

POST/api/v1/scrape200 OK

0.0sFetched example.com/pricingshell only
0.5sRendered in a real browserjs
0.9sBlocked by bot protection403
1.1sEscalated to a residential proxyauto
2.0sPage loaded200
2.3sMarkdown returned1 credit

Example run with proxy: "auto". Timings vary by site.

[ 03 / 07 ]Infrastructure

Infrastructure you don’t have to run.

Browsers, proxies, queues and retries, already running. Long jobs report back through webhooks, so your servers never hold a connection open.

concurrent scrapes at the top tier, unlocked automatically as you spend: 200
pages per crawl job, with depth and path limits: 10,000
scripted browser actions per scrape (click, type, scroll, wait): 25
signed webhook delivery attempts, with backoff: 4×

Keys hashed at rest
API keys are stored as bcrypt hashes.
Private networks blocked
SSRF guards stop requests to internal addresses.
Payments by Stripe
Card details never touch our servers.
30-day job retention
Crawl and extract job data is deleted after 30 days.

[ 04 / 07 ]Agent-native

Built for agents, not only developers.

Agents read the docs, call MCP and wire the API. You add the key.

One prompt. Paste into any coding agent.

Read https://berrycrawl.com/llms.txt and https://docs.berrycrawl.com/docs/guides/agent-quickstart, then integrate Berrycrawl into this repository.

First inspect the repository's language, framework, server entry points, package manager, existing environment files, and test commands. Do not ask me to paste a secret into chat. Tell me to create a Berrycrawl API key in the dashboard and add this variable to the server-side environment file or secret store:

BERRYCRAWL_API_KEY=bc_...

Wait until that variable exists in the environment you will use. Then implement the smallest server-side integration for the feature I asked for. Use the endpoint and request shape from the Berrycrawl docs, keep the key out of browser bundles and source control, add a focused test or local verification path, and report the files changed and the command I can run to verify it.

If the feature starts a crawl or extraction job, persist the returned id and poll GET https://api.berrycrawl.com/api/v1/jobs/{id} until the status is COMPLETED, FAILED, or CANCELLED. Do not create a second job when polling a known id.

Read the agent quickstart

Agent-readable docs. Plain text and JSON at stable URLs.

$ curl -s https://berrycrawl.com/llms.txt

Open llms.txt

[ 05 / 07 ]Developers

Ship it in a few lines of code.

Typed SDKs in nine languages. One OpenAPI contract.

scrape.ts

$ npm i berrycrawl

import { Berrycrawl } from "berrycrawl";

const berrycrawl = new Berrycrawl({
  apiKey: process.env.BERRYCRAWL_API_KEY,
});

const page = await berrycrawl.scrape({
  url: "https://example.com/pricing",
  formats: ["markdown", "links"],
});

200 OK · response

{
  "success": true,
  "data": {
    "markdown": "# Pricing\n\nSimple plans for every team.",
    "links": ["https://example.com/signup"]
  },
  "metadata": {
    "url": "https://example.com/pricing",
    "statusCode": 200
  }
}

Testing? Scrapes and brand lookups of berrycrawl.com are free.

SDK docs API reference OpenAPI

[ 06 / 07 ]Pricing

Pay for the web work you actually run.

No subscription. Paid credits never expire.

Start free

100credits

One-time grant for every new workspace.

No card required
REST, MCP and the dashboard
3 concurrent scrapes to start

Create a workspace

Pay as you go

$1per 1,000 credits

Top up from $10. Limits rise as you spend.

No subscription or minimum term
Failed scrapes cost 0 credits
Up to 200 concurrent scrapes

Get an API key

Credits per call

Credits charged per endpoint
Scrape	1
Parse	1
Crawl / Extract	1 per page
Search	2
Map	2 per 10 URLs
Screenshot	5
YouTube transcript	5
Brand	10

Screenshot, PDF or JSON output on a scrape adds 4 credits; a summary adds 2.

Limits rise automatically with lifetime spend

See full pricing

Free$0
concurrent
3
Tier 1$10
concurrent
10
Tier 2$100
concurrent
30
Tier 3$500
concurrent
80
Tier 4$2,000
concurrent
200

[ 07 / 07 ]FAQ

Questions, answered.

Can't find what you need? The docs cover every endpoint, option and error code.

Read the docs

Product

Does Berrycrawl handle JavaScript-heavy pages?

Yes. Pages that need a browser are rendered in one, and you can script up to 25 browser actions per scrape (click, type, wait, scroll) before the content is captured.

What happens when a site blocks the request?

With the default proxy setting Berrycrawl escalates on its own: direct first, then datacenter, residential and stealth proxies. If a page still comes back as a challenge, CAPTCHA, empty or error page, you get an error code instead of junk content.

Can I crawl a whole website?

Yes. A crawl runs as a background job across up to 10,000 pages with depth and path filters. Poll the job or receive signed webhooks when it starts, completes or fails.

Which file types can /parse read?

Public PDF, Word (DOCX) and Excel (XLSX) files. Each successful parse returns readable markdown for 1 credit.

Agents

How do I connect an AI agent?

Point any MCP client that supports Streamable HTTP at https://api.berrycrawl.com/api/v1/mcp with your key, or paste our setup prompt into your coding agent. Agents can also read /llms.txt for the endpoint map.

Does the agent ever see my API key?

It doesn't need to. The setup prompt asks you to add BERRYCRAWL_API_KEY to the server environment yourself and keeps the key out of chat, browser bundles and source control.

Which SDKs are available?

Typed SDKs for TypeScript, Python, Go, Ruby, PHP, Java, C#, Swift and Rust, all generated from the same OpenAPI contract.

Billing

Do failed requests cost credits?

Blocked, empty and error pages from a scrape cost 0 credits and are never cached. Ordinary pages served from the cache are free too.

Do credits expire?

Paid credits never expire. There is no subscription: top up from $10 when you need more, at $1 per 1,000 credits.

How do rate limits grow?

Limits rise automatically with lifetime spend, from 3 concurrent scrapes on the free tier to 200 at Tier 4. Larger volumes can talk to us about an enterprise tier.

Build the feature. Skip the crawler project.

100 free credits. No card required.

Get an API key Connect your agent

Infrastructure you don’t have to run.

Browsers, proxies, queues and retries, already running. Long jobs report back through webhooks, so your servers never hold a connection open.

concurrent scrapes at the top tier, unlocked automatically as you spend

200

pages per crawl job, with depth and path limits

10,000

scripted browser actions per scrape (click, type, scroll, wait)

signed webhook delivery attempts, with backoff

4×

Read https://berrycrawl.com/llms.txt and https://docs.berrycrawl.com/docs/guides/agent-quickstart, then integrate Berrycrawl into this repository. First inspect the repository's language, framework, server entry points, package manager, existing environment files, and test commands. Do not ask me to paste a secret into chat. Tell me to create a Berrycrawl API key in the dashboard and add this variable to the server-side environment file or secret store: BERRYCRAWL_API_KEY=bc_... Wait until that variable exists in the environment you will use. Then implement the smallest server-side integration for the feature I asked for. Use the endpoint and request shape from the Berrycrawl docs, keep the key out of browser bundles and source control, add a focused test or local verification path, and report the files changed and the command I can run to verify it. If the feature starts a crawl or extraction job, persist the returned id and poll GET https://api.berrycrawl.com/api/v1/jobs/{id} until the status is COMPLETED, FAILED, or CANCELLED. Do not create a second job when polling a known id.

import { Berrycrawl } from "berrycrawl"; const berrycrawl = new Berrycrawl({ apiKey: process.env.BERRYCRAWL_API_KEY, }); const page = await berrycrawl.scrape({ url: "https://example.com/pricing", formats: ["markdown", "links"], });

{ "success": true, "data": { "markdown": "# Pricing\n\nSimple plans for every team.", "links": ["https://example.com/signup"] }, "metadata": { "url": "https://example.com/pricing", "statusCode": 200 } }

Scrape

Parse

Crawl / Extract

1 per page

Map

2 per 10 URLs

Screenshot

YouTube transcript

Brand