One-Liner Scrape, Screenshot, and PDF

Scrape content, take screenshots, or generate PDFs from any URL - without managing sessions, browsers, or cleanup yourself. You provide a URL, you get back the result. The TestMu AI Browser SDK handles everything behind the scenes.

This is useful when your agent needs data from a page but doesn't need to interact with it beyond extraction.

Scrape

Extract content from any webpage in your choice of format.

Simplest form - just pass a URL:

const result = await client.scrape('https://example.com');
console.log(result.content);

With options - control format, timing, and selectors:

const result = await client.scrape({
    url: 'https://example.com',
    format: 'markdown',     // 'html' | 'markdown' | 'text' | 'readability'
    delay: 3000,             // Wait 3 seconds for JS-heavy pages
    waitFor: '#content',     // Wait for this CSS selector before extracting
});

Response:

interface ScrapeResponse {
    title: string;           // Page title
    content: string;         // Extracted content in requested format
    url: string;             // Final URL (after redirects)
    markdown?: string;       // Markdown version
    html?: string;           // Raw HTML
    metadata?: Record<string, string>; // Meta tags
}

Choosing a format:

Format	What You Get	When To Use It
`html`	Raw HTML of the page	When your agent needs to parse the DOM
`text`	Plain text, tags stripped	When you want minimal tokens for LLM input
`readability`	Cleaned article content (like Reader Mode)	When the page has an article you want to extract
`markdown`	HTML converted to markdown	When you want structure + low token count for LLMs

Screenshot

Capture a visual snapshot of any webpage.

// Simple
const result = await client.screenshot('https://example.com');
fs.writeFileSync('screenshot.png', result.data);

// With options
const result = await client.screenshot({
    url: 'https://example.com',
    fullPage: true,          // Capture entire scrollable page
    format: 'jpeg',          // 'png' | 'jpeg' | 'webp'
    quality: 80,             // JPEG/WebP quality (1-100)
    delay: 2000,             // Wait before capturing
});

Response: { data: Buffer, format: string, width: number, height: number }

PDF

Generate a PDF document from any webpage.

// Simple
const result = await client.pdf('https://example.com');
fs.writeFileSync('page.pdf', result.data);

// With options
const result = await client.pdf({
    url: 'https://example.com',
    format: 'A4',            // 'A4' | 'Letter' | 'Legal'
    landscape: false,
    printBackground: true,
    margin: { top: '1cm', right: '1cm', bottom: '1cm', left: '1cm' },
});

Response: { data: Buffer, pageCount: number }

Standalone vs Session Mode

Quick Actions operate in two modes:

Standalone mode (default). The SDK creates a temporary headless browser with stealth enabled, navigates to the URL, performs the operation, and closes the browser. Fully automatic. No session management needed.

Session mode. If your agent already has a session running and you want the Quick Action to use that session's browser (with its cookies, tunnel, or extensions), register the page first:

const session = await client.sessions.create({ ... });
const browser = await client.puppeteer.connect(session);
const page = (await browser.pages())[0];

// Register the page for Quick Actions
client.quick.registerSessionPage(session.id, page);

// Now Quick Actions use the existing session
const result = await client.scrape({
    url: 'https://example.com',
    sessionId: session.id,
});