Skip to main content
POST
Use POST /html to load any public URL in a real browser and retrieve the rendered output in the format your application needs. Choose "html" to get the full DOM, "markdown" to get a clean readable article, or "json" to get a structured Readability payload with title, author, excerpt, and body content. The request is routed through residential proxies with randomized browser fingerprints.
string
required
Target page URL. Only ports 80 and 443 are supported.
string
Controls the response format. Accepted values: "html" (default), "markdown", "json". Both "markdown" and "json" extract readable article content using Readability.
string
Wait for the first matching CSS element before extracting. When contentMode is "html", only that element’s HTML is returned. When "markdown" or "json", Readability runs against that element. Returns 422 CONTENT_SELECTOR_NOT_FOUND if the selector is not found.
string
Navigation event to wait for before extracting content. Defaults to "load".
number
Maximum time to wait for the task to complete, in milliseconds. Must be a positive integer no greater than 3,600,000. Values outside this range fall back to the server default.
object
Viewport dimensions used when rendering the page.
string
Browser locale, for example "en-US" or "zh-CN". Affects navigator.language and Accept-Language headers.
string
IANA timezone identifier, for example "Asia/Shanghai" or "America/New_York".
object
Geolocation coordinates exposed to the page via the Geolocation API.
object
Custom proxy configuration. Cannot be combined with countryCode. Provide either server or the split form (protocol + host + port). Credentials must not be embedded in server.
string
Managed residential proxy region. Cannot be combined with proxy.
  • "GLOBAL": dynamic exit from 15 popular regions.
  • Two-letter country code (e.g. "US", "DE"): prefers a trusted proxy for that region with dynamic fallback.
  • Omitted: a random trusted proxy is selected automatically.
string
"random" lets the server pick a User-Agent from its library. Requests without an explicit userAgent already default to "random".
string
Operating system used when userAgentMode is "random". Accepted values: "windows" (default) or "macos".
string
Explicit User-Agent string. Overrides the random selection.
object
Browser fingerprint settings. When omitted, every signal defaults to random while keeping OS, GPU, CPU, memory, fonts, and device signals coherent.
array
Cookie list injected into the browser context before navigation. Each cookie object requires name, value, and domain.

Response

string
Article title extracted from the page.
string
Author name or byline, if present.
string
A short summary or description of the article.
string
Name of the site, if available.
string
Language code of the article, e.g. "en".
string
Text direction, e.g. "ltr" or "rtl". null if not detected.
string
Cleaned HTML of the article body.
string
Plain-text version of the article body with whitespace normalised.
number
Character count of textContent.
string
ISO 8601 publication timestamp, or null if not found.
One credit is consumed after request validation but before the task is enqueued. Request bodies are limited to 1 MiB.

More examples