Skip to content

meow:agent-browser

Chrome/Chromium browser automation via CDP for page navigation, form filling, screenshots, session persistence, and auth flows.

What This Skill Does

meow:agent-browser wraps the agent-browser CLI (vercel-labs) to control Chrome/Chromium directly via Chrome DevTools Protocol. Its key advantage over other browser skills is session persistence — it can import auth from your real browser, save/restore session state, and use persistent profiles. This makes it the best choice for authenticated flows.

Core Capabilities

  • CDP-based control — Direct Chrome DevTools Protocol, not Playwright wrapper
  • Session persistence — Profiles, session names, state files for auth reuse
  • Auth import — Grab cookies from a running Chrome instance (--auto-connect)
  • Element references — Snapshot with @e1, @e2 refs for precise interaction
  • Video recording — Record browser sessions for debugging
  • Proxy support — Route through HTTP/SOCKS proxies

When to Use This

Use agent-browser when...

  • You need authenticated browser sessions that persist
  • You need to import auth from your real browser
  • You're testing flows that require complex session management
  • You need video recording of browser interactions

For simpler browser tasks...

Usage

bash
# Navigate and interact
agent-browser open https://example.com
agent-browser snapshot -i          # see elements with @e1, @e2 refs
agent-browser fill @e1 "user@example.com"
agent-browser click @e3

# Import auth from your running Chrome
agent-browser --auto-connect state save ./auth.json
agent-browser --state ./auth.json open https://app.example.com/dashboard

# Persistent profile (stays logged in across sessions)
agent-browser --profile ~/.myapp open https://app.example.com

Example Prompts

PromptWhat happens
open the dashboard and take a screenshotNavigate → wait for load → screenshot
log in with my browser sessionImport auth from running Chrome → navigate
test the checkout flowNavigate → fill form → submit → verify

Quick Workflow

agent-browser open [url]
  → agent-browser snapshot -i (get element refs)
  → agent-browser fill/click (interact using refs)
  → agent-browser snapshot (verify result)

Skill Details

Phase: 2–4

Gotchas

  • Stale refs after dynamic DOM updates: Modals, infinite scroll, and tab switches all invalidate refs silently — commands succeed but target the wrong element. Re-run snapshot -i after any interaction that causes DOM change, not just navigation.
  • Cross-origin iframes block CDP: Sandboxed iframes (Stripe, reCAPTCHA) appear in snapshot but fill/click fail silently. Use screenshot --annotate to confirm reachability; use --auto-connect against a browser where user has already interacted.
  • JavaScript dialogs freeze all commands: An unhandled alert()/confirm()/prompt() times out every subsequent command. Run agent-browser dialog status first when debugging unexpected timeouts; dismiss with dialog accept or dismiss.

Released under the MIT License.