SDK quick actions reference

Quick actions create an ephemeral browser for one deterministic operation. They are useful when you do not need a persistent browser handle or a model-backed agent.

const page = await web.scrape({
  url: "https://www.sec.gov/edgar/search/",
  format: "markdown",
  maxChars: 20_000,
  policy: { allowedDomains: ["sec.gov"] },
});

All quick actions accept policy. All quick actions also accept proxy, which routes the ephemeral browser through a custom proxy URL.

`web.scrape()`

web.scrape(options: string | ScrapeOptions): Promise<ScrapeResponse>

Pass a URL string for the default markdown read:

const result = await web.scrape("https://example.com");

Use ScrapeOptions for control:

Field	Type	Default	Description
`url`	`string`	Required	URL to read.
`format`	`"html" \| "markdown" \| "text"`	`markdown`	Returned content format.
`selector`	`string`	None	CSS selector that scopes page content.
`iframe`	`string`	None	CSS selector of a specific iframe to target.
`waitFor`	`string`	None	CSS selector to wait for before reading.
`timeout`	`number`	`30000`	Total operation timeout in milliseconds.
`maxChars`	`number`	`20000`	Soft cap on returned content characters. Maximum is `50000`.
`startFromChar`	`number`	`0`	Zero-based content character offset for continuing a truncated read.
`readFrom`	`"start" \| "end"`	`start`	Read window direction.
`includeLinks`	`boolean`	`false`	Include deterministic anchor links.
`iframeText`	`boolean`	`false`	When `format` is not `html`, append iframe text to content.
`proxy`	`string`	None	Proxy URL for the ephemeral browser.
`policy`	`BrowserNavigationPolicy`	Client-level policy, if configured	Browser navigation policy for the read.

ScrapeResponse:

Field	Description
`v`	Response version. Current value is `1`.
`url`	Final URL.
`title`	Page title.
`format`	Returned format.
`content`	Primary deterministic read content.
`provenance`	Read source and scoping metadata.
`diagnostics`	Read threshold and fallback diagnostics.
`completeness`	Ratio and cap/pruning diagnostics.
`statusCode`	Page HTTP status code.
`links`	Deterministic links when `includeLinks` is true.
`truncated`	Whether `content` was clamped to `maxChars`.
`charsOmitted`	Omitted UTF-16 code units when truncated.
`warnings`	Structured warnings.
`tips`	Browser policy tips.
`elapsedMs`	Browser-side elapsed time.

`web.screenshot()`

web.screenshot(options: ScreenshotOptions): Promise<ScreenshotResponse>

const shot = await web.screenshot({
  url: "https://example.com/dashboard",
  fullPage: true,
  format: "png",
});

ScreenshotOptions:

Field	Type	Default	Description
`url`	`string`	Required	URL to capture.
`fullPage`	`boolean`	`false`	Capture the full scrollable page.
`format`	`"png" \| "jpeg" \| "webp"`	`png`	Image format.
`quality`	`number`	`80`	Image quality for `jpeg` and `webp`.
`selector`	`string`	None	CSS selector of a specific element to capture.
`hideSelectors`	`string[]`	None	CSS selectors hidden with `visibility:hidden` during capture.
`maskSelectors`	`string[]`	None	CSS selectors covered by screenshot masks.
`maskColor`	`string`	None	Mask color.
`style`	`string`	None	Temporary screenshot-only CSS.
`waitFor`	`string`	None	CSS selector to wait for before capture.
`timeout`	`number`	`30000`	Total operation timeout in milliseconds.
`proxy`	`string`	None	Proxy URL for the ephemeral browser.
`policy`	`BrowserNavigationPolicy`	Client-level policy, if configured	Browser navigation policy for capture.

ScreenshotResponse:

Field	Description
`url`	Final URL.
`data`	Base64-encoded image data.
`format`	Image format.
`width`	Image width.
`height`	Image height.
`elapsedMs`	Browser-side elapsed time.
`artifact`	Sensitivity and export metadata.
`tips`	Browser policy tips.
`_meta`	Quick-action metadata.

`web.pdf()`

web.pdf(options: PdfOptions): Promise<PdfResponse>

const pdf = await web.pdf({
  url: "https://example.com/report",
  format: "Letter",
  landscape: false,
});

PdfOptions:

Field	Type	Default	Description
`url`	`string`	Required	URL to render.
`format`	`"A4" \| "Letter" \| "A3" \| "Tabloid" \| "Ledger" \| "Legal"`	`A4`	Paper size.
`landscape`	`boolean`	`false`	Use landscape orientation.
`printBackground`	`boolean`	`true`	Print background graphics.
`waitFor`	`string`	None	CSS selector to wait for before PDF generation.
`timeout`	`number`	`30000`	Total operation timeout in milliseconds.
`proxy`	`string`	None	Proxy URL for the ephemeral browser.
`policy`	`BrowserNavigationPolicy`	Client-level policy, if configured	Browser navigation policy for rendering.

PdfResponse:

Field	Description
`url`	Final URL.
`data`	Base64-encoded PDF data.
`pages`	Page count.
`elapsedMs`	Browser-side elapsed time.
`tips`	Browser policy tips.
`_meta`	Quick-action metadata.

Proxy example

const result = await web.scrape({
  url: "https://example.com",
  proxy: process.env.RESIDENTIAL_PROXY_URL,
});

Proxy credentials can appear inside URLs. Keep proxy URLs in environment variables and avoid logging request objects directly.

Get started

Build agent workflows

Observe and debug

Runtime concepts

Examples

Integrations

Production

Reference

API reference

SDK quick actions reference

`web.scrape()`

`web.screenshot()`

`web.pdf()`

Proxy example

​web.scrape()

​web.screenshot()

​web.pdf()

​Proxy example

`web.scrape()`

`web.screenshot()`

`web.pdf()`

Proxy example