Skip to main content
Every browser runs in a hardened Chromium fork with stealth, anti-fingerprinting, and residential proxies enabled by default.
This page is for direct browser control. To give an AI agent a goal instead, create an API V4 run.

1. Create a browser

2. Connect over CDP

Playwright

Puppeteer

Selenium

Selenium’s debugger_address only supports local host:port connections. Use Playwright or Puppeteer for remote CDP over WebSocket.

3. Stop the browser

browser.close() and disconnecting CDP do not stop the managed browser. Call PATCH /api/v4/browsers/{id} with {"action":"stop"}.
See Create browser session and Update browser session for every browser setting and response field.

PDF rendering

Standalone browser creation accepts "pdfRendererEnabled": false to disable Chrome’s in-tab PDF viewer. PDFs are still saved to the browser session’s download directory. This does not block PDF downloads or reduce their bytes to zero. The setting is not currently exposed in V4 agent browserSettings.

Profiles and browser contexts

To reuse a saved login, supply profileId when creating the browser and use the existing browser context shown above. Creating a fresh incognito context with new_context() / newContext() does not inherit that profile’s cookies. A profile is not a guarantee that the site’s login remains valid, and cookie sync does not import saved passwords or promise a persistent HTTP disk cache. See Sync local and cloud cookies.