cloudflare-browser-rendering

Screenshots on Cloudflare Workers use headless Chromium via Puppeteer or Playwright bindings. Set up a browser binding in wrangler.jsonc, install the library, then call page.screenshot() to capture and return images at the edge.

Yes. Cloudflare provides @cloudflare/puppeteer and @cloudflare/playwright packages that run headless Chrome natively on Workers. Both support screenshots, PDFs, web scraping, and browser automation without external APIs.

Manage session limits through worker bindings configuration and session history tracking. Use incognito contexts, implement retry logic for timeouts, call disconnect() to release sessions, and monitor binding-not-passed errors to stay within cost limits.

Use Puppeteer or Playwright's multi-page automation to launch or connect sessions, navigate pages sequentially, and extract data. Apply session management patterns and error handling with retries to handle XPath errors and maintain stability across crawls.

Yes. Puppeteer and Playwright on Workers both support PDF generation via page.pdf(). Render the page content, configure PDF options, and return the buffer—all executed at the edge without external services.

Binding-not-passed errors occur when the browser binding isn't declared in wrangler.jsonc or the Worker doesn't have permission to access it. Verify the binding configuration, ensure the Worker is deployed with the correct service binding, and check that launch or connect calls reference the binding correctly.