Browsers

Some pages are built by JavaScript after they load: the HTML the server sends doesn't contain the data. To get it, load the page in a real browser with options.browser:

"options": { "browser": true }

See it workCopy Anchor Link

This page writes a paragraph with JavaScript. Without a browser, the API reads the HTML as sent, before any script runs:

With a browser, the script runs first:

{
  "browser": { "engine": "chromium", "driver": "playwright", "stealth": "plugin" },
  "data": ["Served by the server", "Rendered by JavaScript"]
}

A browser is slower than a plain request (about a second, more for stealth engines) and uses more resources. Use it when the data needs JavaScript, or when a site only lets real browsers in.

EnginesCopy Anchor Link

Pick one by name:

"options": { "browser": "camoufox" }

GET /browsers lists the engines installed on this server.

DriversCopy Anchor Link

A driver is the library that controls the browser. Each engine has a default; choose another with driver:

"options": { "browser": { "engine": "chromium", "driver": "puppeteer" } }

StealthCopy Anchor Link

stealth is on by default. What it does depends on the engine:

"stealth" in the response tells you what applied: "plugin", "builtin" or "off".

OptionsCopy Anchor Link

browser accepts true, an engine name, or an object:

"options": {
  "browser": {
    "engine": "cloakbrowser",
    "wait_until": "networkidle",
    "wait_for": ".product",
    "block": ["image", "font"]
  }
}

Block resourcesCopy Anchor Link

Pages load faster and use less proxy bandwidth when the browser skips what scraping doesn't need. block lists what to skip:

The default is ["media", "font", "ads", "trackers"]. Your list replaces it, it doesn't add to it, and false loads everything:

"browser": { "block": ["image", "media", "font", "stylesheet", "ads", "trackers"] }

Ads and trackers are refused by host name, whatever the engine and driver. If the page you scrape is itself on one of those hosts, it still loads. Without a browser, block has nothing to do: only the HTML is downloaded.

Capture the page's API callsCopy Anchor Link

Many sites load their data from their own JSON API after the page loads. Instead of scraping the rendered HTML, capture those responses as they are:

"captured": [
  { "url": "…/demo/shop/api", "status": 200, "json": { "page": 1, "stock": [ … ] } }
]

Only JSON responses are kept, 20 at most.

Everything else still appliesCopy Anchor Link

Browsers work with the other options:

• Actions click, type and scroll in the page.
• Proxies, of any type, carry the browser's traffic.
• Custom requests: method, body, headers and cookies apply to the first request, the one that loads the page.
• Retries, fallbacks, timeouts and the cache work the same.
• Cookies set by the site are kept for the rest of the request: actions, retries and fallbacks, browser or not.

When something can't be doneCopy Anchor Link

Rather than failing, the API adapts the request and says so in warnings:

"warnings": ["camoufox can't be driven by puppeteer: used playwright instead"]

Set "strict": true to get 503 BROWSER_UNAVAILABLE instead of a substitution.