Features / Automation & API

Everything the app does, from your own code.

Browz runs a local API on your computer with about 190 endpoints, a command-line tool, and an MCP server for AI agents. Start a profile and connect Playwright or Puppeteer to it, or record a flow and run it on every profile.

Join the beta See all features

Start a profile# start by name or id browz start "Shop – Sydney 01" # or call the local API curl -X POST http://127.0.0.1:50333/api/v1/browser/start \ -H "Authorization: Bearer $BROWZ_TOKEN" \ -d '{"id":"pe44303bd43d3fb2a"}'

Drive it with Playwrightimport { chromium } from 'playwright'; const browser = await chromium.connectOverCDP( `http://127.0.0.1:${debugPort}`); const page = browser.contexts()[0].pages()[0]; await page.goto('https://example.com');

A local API for everything.

The same API the app uses, on 127.0.0.1 only, protected by a bearer token you can rotate from the app. Create profiles, manage proxies, start and stop browsers, read fingerprints and run bulk jobs. A built-in playground lets you try any endpoint.

About 190 endpoints under /api/v1
Loopback only, with a rotating token
Try-it playground in the app

Browz · API & MCP

Method	Endpoint	Does
POST	/api/v1/profile/create	Create a profile
POST	/api/v1/browser/start	Start a browser
POST	/api/v1/browser/stop	Stop it, keep the session
POST	/api/v1/proxy/create	Add a proxy
GET	/api/v1/proxy/list	The proxy pool

A command line that reads like English.

The browz command lists, creates, starts, stops and updates profiles by name or id, manages proxies and runs checks, and makes raw API calls when you need them. Add --json for scripts.

list, start, stop, create, update, delete
proxy add, set, rm and check
import csv, bulk jobs and raw json calls

Terminalbrowz create "Shop – Sydney" --count 5 \ --group "Shops / AU" --proxy "Sydney residential" # → Shop – Sydney-1 … Shop – Sydney-5 browz --json list | jq '.[].name'

Give an AI agent the keys, safely.

The bundled MCP server lets an AI assistant list, create, start and stop profiles, navigate pages, take screenshots, evaluate scripts and export cookies, all through the same token-protected local API.

Profiles, pages, screenshots and cookies
Runs locally, next to the app
Works with any MCP client

MCP client config{ "mcpServers": { "browz": { "command": "/Applications/Browz.app/Contents/Resources/browz-mcp" } } }

Record a flow once. Run it on every profile.

Build a flow from steps: go to a page, wait, click, type, scroll, extract a value into a variable, run a script, branch on whether an element exists, or loop. Test it on one profile, then run it on every running profile in turn.

Nine step types, with variables and branches
Run on one profile or all running profiles
Run history; typed text is never logged

Browz · RPA

Flow · Check order status

Ran on 4 profiles

Go tohttps://example.com/ordersgoto

Wait for.order-listwait

Extract.status → {{status}}extract

If.alert existsif/else

FAQ

About automation.

Does the API need an internet connection or a cloud account?

The API itself runs on your computer, on the loopback interface only; your scripts never go through a cloud API. Browz is cloud-only, though: you sign in to your Browz account, and starting a profile needs your account server, which keeps a profile open on one device at a time.

Is the API compatible with other tools’ APIs?

No. Browz has its own API. Connecting Playwright, Puppeteer or Selenium over the Chrome DevTools Protocol works the same way as with any Chromium browser.

Which MCP clients work?

Any client that runs a local MCP server over stdio.

Read the developer quickstart

Private beta

Every account. Its own browser.

Join the beta See pricing

For Mac and Windows.

Everything the app does, from your own code.

A local API for everything.

A command line that reads like English.

Give an AI agent the keys, safely.

Record a flow once. Run it on every profile.

Flow · Check order status

About automation.

Works well with

Browser profiles

Proxies

Synchronizer

Every account. Its own browser.