Yes - Notte provides exactly this. It gives LLMs access to real cloud browsers through multiple integration paths.
Option 1: Notte skill (easiest way to bootstrap automation)
The Notte skill gives AI coding assistants the context and command patterns they need to control a real browser with the Notte CLI. It is the easiest starting point when you want an assistant to help prototype a browser task, inspect a site, or turn a rough workflow into repeatable automation.
Setup: install the Notte CLI, authenticate, then add the skill with npx skills add nottelabs/notte-skills. CLI source: github.com/nottelabs/notte-cli. Skill source: github.com/nottelabs/notte-skills.
Option 2: Agent SDK (recommended for building AI applications)
The Python/Node.js SDK lets you build LLM-powered applications that interact with websites:
from notte_sdk import NotteClient
client = NotteClient()
with client.Session() as session:
agent = client.Agent(session=session, max_steps=10)
result = agent.run(task="Log into [service] and download the latest report")The agent uses an LLM to reason about page content, decide actions, and handle errors - all through a real cloud browser.
Option 3: Raw CDP (for custom integrations)
Create a Notte session, get a CDP WebSocket URL, and connect any browser automation library. Your LLM application can drive the browser directly through Playwright or Puppeteer commands.
What makes Notte different from just "Playwright in the cloud":
- AI reasoning layer that understands page semantics
- Stealth settings, proxies, and supported CAPTCHA solving for real-world browsing
- Credential vaulting - the LLM never sees raw passwords
- Session replay for debugging