#Browser (CDP)

When an agent needs full page automation, hand the sandbox's Chromium to Playwright or Puppeteer. The daemon supplies the CDP endpoint, and the client connects to and drives the same browser instance.

Routes:

For automation without a framework — navigate, snapshot, fill, click, screenshot over REST — see Browser Use.

#Requirements

A Chromium started with --remote-debugging-port=9222 must be reachable from the daemon. The daemon connects to it as a CDP client; it does not launch or proxy it.

#Get the CDP URL

cdp_url is Chromium's WebSocket endpoint rewritten to the public origin:

BASE_URL = "http://127.0.0.1:18091"
sb = Aio(BASE_URL)
cdp_url = sb.get("/v2/browser/info")["cdp_url"]
print(cdp_url)
# -> ws://127.0.0.1:18091/cdp/devtools/browser/9ca46b6e-ab32-46b1-8e7e-7af3d540de4b

BASE_URL = "http://127.0.0.1:18091"
sb = Aio(BASE_URL)
# The 1.x Python SDK's browser.get_info rejects this body; see the migration
# guide. The route itself answers, so the helper reads it.
cdp_url = sb.get("/v1/browser/info")["cdp_url"]
print(cdp_url)
# -> ws://127.0.0.1:18091/cdp/devtools/browser/9ca46b6e-ab32-46b1-8e7e-7af3d540de4b

The path lives under /cdp/, which the deployment serves, not the daemon. See CDP Access. Which SDK calls fail against the daemon is listed in 1.x SDK compatibility.

#Connect with Playwright

Pass cdp_url to connect_over_cdp:

import httpx
from playwright.sync_api import sync_playwright

BASE_URL = "http://127.0.0.1:18091"
cdp_url = httpx.get(f"{BASE_URL}/v1/browser/info").json()["data"]["cdp_url"]

with sync_playwright() as p:
    browser = p.chromium.connect_over_cdp(cdp_url)
    page = browser.contexts[0].pages[0]
    page.goto("https://example.com")
    print(page.title())
    # → Example Domain

import { chromium } from "playwright-core";

const BASE_URL = "http://127.0.0.1:18091";
const info = await fetch(`${BASE_URL}/v1/browser/info`).then((r) => r.json());

const browser = await chromium.connectOverCDP(info.data.cdp_url);
const page = browser.contexts()[0].pages()[0];
await page.goto("https://example.com");
console.log(await page.title());
// → Example Domain

Playwright keeps the /cdp/ prefix and any query string, so cdp_url works unmodified. Take the existing context and page rather than launching one: the browser belongs to the sandbox and is shared with the daemon's own REST tools.

#Connect with Puppeteer

Use browserWSEndpoint. Puppeteer's browserURL drops the path prefix:

import puppeteer from "puppeteer-core";

const BASE_URL = "http://127.0.0.1:18091";

const info = await fetch(`${BASE_URL}/v1/browser/info`).then((r) => r.json());
const browser = await puppeteer.connect({ browserWSEndpoint: info.data.cdp_url });
const [page] = await browser.pages();
await page.goto("https://example.com");
console.log(await page.title());
// → Example Domain

#Related

• Browser Use — the same Chromium driven by REST calls alone
• Browser API — the full REST tool list, MCP tools, and SDK notes
• CDP Access — how a deployment exposes /cdp/*
• Migration: Removed routes — the page automation routes CDP replaces