Skip to main content
AgentsUse

Browser Automation

Browserbase

A cloud platform that gives agents browsers to drive, plus search, fetch, and a runtime for running them.

Official siteDocumentation
usage_basedChecked 2026-10-07Source verified

Quick decision

Best for
Agents that need real browsers without operating a fleet Teams already using Stagehand for agent browser control Workflows mixing browser driving with search and fetch APIs on one bill
Not ideal for
Self-hosted or offline requirements Simple URL fetching where a reader API is cheaper and simpler
Pricing model
Hosted platform billed by usage; current plans and session pricing are published on the Browserbase site.
Deployment
cloud.
Authentication
api_key
Review state
Source verified. Facts checked 2026-10-07. Not locally tested by AgentsUse.

What it does

One API key gives your agent access to cloud browsers, web search, page fetching, and a sandbox runtime to run it all, in Browserbase own summary of the platform. The center of the product is cloud browser sessions: your agent connects to a real browser without running Chromium yourself, and Browserbase handles the surrounding fleet work.

Around the browsers sit three supporting APIs. The Search API looks things up on the web, the Fetch API retrieves pages as usable content, and Functions and the agent runtime provide a place to run the agent code itself. Stagehand, Browserbase open-source SDK, is the usual way agents drive these browsers with natural-language actions.

Browserbase is a hosted platform, so availability, regions, and abuse handling are the vendor problem, while per-session usage is yours. Teams should expect to wire observability (session replays and logs) into their own failure review rather than treating cloud browsers as invisible.

Verified capabilities

Core capability

  • Cloud browsers Source verified

    Hosted browser sessions an agent can drive, with the fleet operated by Browserbase.

  • Search API Source verified

    A web search API the agent can call alongside its browser sessions.

  • Fetch API Source verified

    Retrieves pages as content an agent can read, without spinning up a full browser for simple fetches.

  • Agent runtime Source verified

    A sandbox runtime to run agent code next to the browsers and APIs.

  • Stagehand SDK integration Source verified

    The open-source Stagehand SDK is the usual control layer for Browserbase sessions, mixing code and natural-language actions.

Quick start

Connect to a cloud browser

Browserbase returns a CDP endpoint for each cloud session.

Connect to a cloud browser
npm install @browserbasehq/sdk puppeteer-core

import Browserbase from "@browserbasehq/sdk";
import puppeteer from "puppeteer-core";

const bb = new Browserbase({ apiKey: process.env.BROWSERBASE_API_KEY });
const session = await bb.sessions.create({ projectId: process.env.BROWSERBASE_PROJECT_ID });
const browser = await puppeteer.connect({ browserWSEndpoint: session.connectUrl });

Source: https://docs.browserbase.com/introduction. Examples use placeholders only. Never paste a real key into a profile, config file you share, or a ticket.

MCP support: Browserbase tooling is reachable from MCP-capable agents through its SDKs and Stagehand; check the official docs for the current MCP setup.

Works with

Frameworks

Stagehand, LangChain, CrewAI, Any CDP-capable client

SDKs and packages

TypeScript, Python

Works with

Puppeteer, Playwright, Selenium, Chrome DevTools Protocol clients

Only sourced support is listed. A missing framework means AgentsUse has not verified it yet, not that it cannot work.

Health and maintenance

License
usage_based
Maintainer
Browserbase

Closed hosted platform; profile facts read from the official Browserbase documentation introduction on 2026-10-07. No public repository metric is claimed for the platform itself.

Pricing and license

Hosted platform billed by usage; current plans and session pricing are published on the Browserbase site.

Official pricing or docs →

Limitations and safety

  • Hosted only: your agents depend on Browserbase availability and session policies.
  • Usage billing means long browsing tasks need session time monitoring.
Safety

Browser and data tools can read pages, fill forms, and download files. Start with a test account or read-only access, keep credentials in environment variables, and review agent actions before connecting anything that can spend money, send messages, or delete data.

Alternatives to Browserbase

Steel

Open-source browser API you can self-host, with a hosted cloud from the same project

Tradeoff: Its profile lists its own limits, pricing, and verification details for a side-by-side read.

Playwright

Run your own browsers with the automation library directly

Tradeoff: Its profile lists its own limits, pricing, and verification details for a side-by-side read.

Stagehand

The agent control SDK alone, on browsers you run or host elsewhere

Tradeoff: Its profile lists its own limits, pricing, and verification details for a side-by-side read.

Common questions

What does Browserbase do for an AI agent?

A cloud platform that gives agents browsers to drive, plus search, fetch, and a runtime for running them.

Is Browserbase open source?

This profile records the pricing model as usage_based. See the pricing section for the sourced summary.

Does Browserbase support MCP?

Browserbase tooling is reachable from MCP-capable agents through its SDKs and Stagehand; check the official docs for the current MCP setup.

Sources and freshness

Last checked 2026-10-07. Verification label: source verified, which means public claims trace to the sources above. It does not mean AgentsUse ran the tool. Spotted an error? Send a correction. Back to the tools directory.