agent-browser 101: AI-First Browser Automation CLI

Install, basic commands, Accessibility Tree ref-based element selection, and proxy configuration for scraping.

16Yun Engineering TeamMay 3, 20261 min read

Installation

npm install -g agent-browser
agent-browser install  # Auto-downloads Chrome for Testing on first run

macOS Homebrew

brew install agent-browser
agent-browser install

Verify

agent-browser doctor

Fastest Browser Automation Workflow

# 1. Open a page
agent-browser open https://example.com

# 2. Get page structure (Accessibility Tree)
agent-browser snapshot

# 3. Operate elements by ref
agent-browser click @e2
agent-browser fill @e3 "user@example.com"

# 4. Screenshot
agent-browser screenshot page.png

# 5. Close
agent-browser close

Zero code. All CLI.

Snapshot + Ref Selection

The key difference from traditional browser automation tools:

MethodToolPrincipleStability
CSS selectorPlaywright/PuppeteerDOM classes/IDsBreaks on redesign
XPathSeleniumDOM pathBreaks on structure change
Accessibility Refagent-browserAccessibility treeStable if semantics stay

Workflow

# Get snapshot with refs
agent-browser snapshot
# [1] @e1 heading "Welcome"
# [2] @e2 button "Sign In"
# [3] @e3 textbox "Email"
# [4] @e4 textbox "Password"
# [5] @e5 button "Submit"

# Operate by ref
agent-browser click @e2
agent-browser fill @e3 "my@email.com"
agent-browser fill @e4 "mypassword"
agent-browser click @e5

Traditional Selectors Also Work

agent-browser click "#submit"
agent-browser find role button click --name "Submit"

Quick Command Reference

CategoryCommandDescription
Navigateopen <url>Open a URL
ExploresnapshotGet Accessibility Tree + refs
Clickclick <sel>Click an element
Fillfill <sel> <text>Clear and fill
Typetype <sel> <text>Append text
Screenshotscreenshot [path]Take screenshot
Screenshotscreenshot --fullFull page screenshot
Screenshotscreenshot --annotateAnnotated with numbered labels
JSeval <js>Execute JavaScript
Getget title / get urlGet title/URL
Getget text <sel>Get element text
Waitwait <ms> / wait <sel>Wait
ClosecloseClose browser

Using with Proxies

export HTTP_PROXY=http://user:pass@proxy.16yun.cn:8888
export HTTPS_PROXY=http://user:pass@proxy.16yun.cn:8888

agent-browser open https://example.com

Verify Proxy

agent-browser open https://httpbin.org/ip
agent-browser get text body
ScenarioProxy ProductSetup
Single testCrawler ProxyHTTP_PROXY env var
Batch scrapingAPI ProxySwitch proxy per session
Long-runningDedicated ProxyFixed IP + session persistence

CLI + AI Agent Workflow

agent-browser batch \
  "open https://example.com/login" \
  "fill @e3 admin@example.com" \
  "fill @e4 password123" \
  "click @e5"

# Natural language control
agent-browser chat "open example.com, log in, and screenshot the dashboard"

Need an enterprise proxy plan?

We can tailor architecture to your target domains, concurrency, and reliability goals.