Skip to main content

Command Reference

Complete index of Browser-act CLI commands.

Written by Maggie

This page is a command index. Use the feature chapters for context, examples, and decision rules.

Global options

Note: Global flags can be combined with command groups. Use --session <name> for browser interaction commands, --format text|json for output, --no-auto-dialog when dialogs should not be handled automatically, and --version to check the installed CLI.

Option

Description

--session <name>

Session name for browser interaction commands

--format text|json

Output format

--no-auto-dialog

Disable automatic dialog handling

--version

Show CLI version

Goal-based lookup

I want to...

Use

Read more

Extract protected page content

stealth-extract

Anti-detection & Blocking

Create a stealth browser

browser create --type stealth ...

Browser Modes

Open a page and interact

browser open -> state -> click / input

Quick Start

Find API calls behind the UI

network requests -> network request <request_id>

Designed for Agents

Handle CAPTCHA

captcha-aid, solve-captcha, remote-assist

Anti-detection & Blocking

Let a human take over

remote-assist --objective "..."

Remote Assist

Run multiple accounts

separate browsers and named sessions

Concurrency & Isolation

Reuse local login state

browser import-profile

Browser Modes

Command groups

Browser interaction

Navigate pages, read state, click, type, extract, wait, capture network data, and manage tabs.

Browser management

List, create, update, delete, open, and import browser profiles.

Sessions

List and close task workspaces.

Stealth extraction

Extract protected content without opening a named session.

Authentication

Log in, poll device auth, set API keys, or clear credentials.

Skill and system

Load runtime-aware instructions, send feedback, and report logs.

Browser interaction commands

These commands require --session <name>.

Navigation

navigate <url> [--new-tab]
back
forward
reload

Page state

state
screenshot [path] [--full]

state uses indexed markers:

  • [N]: element index for commands such as click N or input N "text"

  • *[N]: element is new or changed since the previous state call

Interaction

click <index>
hover <index>
input <index> <text>
select <index> <option>
type <text>
keys <key_combo>
scroll up|down [--amount]
scrollintoview --selector <css>
upload <index> <path>

Data extraction

get title
get html [--selector]
get text <index>
get value <index>
get markdown

JavaScript

eval <js> [--stdin]

Use eval for complex extraction, page-local calculations, or fallback operations that are awkward through indexed interaction.

Waiting

wait stable [--timeout]
wait selector <index> --state visible|hidden|attached|detached [--timeout]
wait selector --selector <css> --state visible|hidden|attached|detached [--timeout]

Network

network requests [--filter] [--type] [--method] [--status] [--clear]
network request <request_id>
network clear
network offline on|off
network har start
network har stop [path]

Use network capture to inspect XHR requests, find API endpoints, debug authentication, or export HAR files.

Tabs and dialogs

tab list
tab switch <tab_id>
tab close [tab_id]
dialog accept [text]
dialog dismiss
dialog status

Cookies

cookies get [--url]
cookies set <name> <value> [--domain] [--path] [--secure] [--http-only]
cookies clear [--url]
cookies export <file>
cookies import <file>

CAPTCHA and handoff

captcha-aid
solve-captcha
remote-assist [--objective]

Command

Purpose

captcha-aid

Assist with CAPTCHA on the current page

solve-captcha

Try automatic CAPTCHA solving on the current page

remote-assist

Generate a live URL so a human can operate the browser

Browser management commands

These commands do not require a session unless noted by the command behavior.

browser list
browser open <id_or_url> [url_arg] [--headed] [--private] [--allow-restart-chrome]
browser create --type <chrome|chrome-direct|stealth> --name <name> --desc <description> [options]
browser update <browser_id> [options]
browser delete <browser_id>
browser regions
browser list-profiles
browser import-profile <browser_id> <profile_id> [--allow-restart-chrome]

browser create options

browser-act browser create \
  --type chrome|chrome-direct|stealth \
  --name "my-browser" \
  --desc "Purpose description" \
  --source-profile <profile_id> \
  --dynamic-proxy <region> \
  --custom-proxy <url> \
  --private true|false \
  --confirm-before-use

Option notes:

  • --source-profile applies to chrome browsers.

  • --dynamic-proxy, --custom-proxy, and --private apply to stealth browsers.

  • --confirm-before-use marks a browser as sensitive so the agent should ask before opening it.

browser update options

browser-act browser update <browser_id> \
  --name "New name" \
  --desc "Replace the description" \
  --desc-append "Append to the description" \
  --dynamic-proxy <region> \
  --custom-proxy <url> \
  --no-proxy \
  --private true|false \
  --confirm-before-use|--no-confirm-before-use

Session management

session list
session close [name]

Stealth extraction

stealth-extract is independent. It does not require a session.

browser-act stealth-extract <url>
browser-act stealth-extract <url> --content-type html|markdown
browser-act stealth-extract <url> --dynamic-proxy <region>
browser-act stealth-extract <url> --custom-proxy <proxy_url>
browser-act stealth-extract <url> --timeout 60
browser-act stealth-extract <url> --output ./result.md

Authentication

auth login
auth poll
auth set <api_key>
auth clear

Skill and system

get-skills core --skill-version <version>
get-skills advanced
get-skills main
report-log
feedback <message>

Learn more

Designed for Agents

Learn why the command model is optimized for agents.

Quick Start

Combine commands into the first working browser loop.

Did this answer your question?