agent-browser

Automate Chrome browser interactions via CDP with snapshots, frames, and sessions.

1|3|Updated Apr 11, 2026
One-click install
npx skills add https://github.com/kunmath/split-it --skill agent-browser-kunmath
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/kunmath/split-it/tree/main/.agents/skills/agent-browser
Command: npx skills add https://github.com/kunmath/split-it --skill agent-browser-kunmath

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Browser automation for AI agents is tedious and error-prone when manually scripting interactions with websites. agent-browser provides a CLI to drive Chrome/Chromium via CDP, enabling deterministic actions, frame handling, authentication, and session management in a single tool. This Skill enables rapid remote automation, testing, data extraction, and task automation from natural language prompts.

Core Features & Use Cases

  • Snapshot and refs for low-context ML guidance
  • Interact with elements via @refs to perform clicks, fills, and navigation
  • Session management with named sessions, state persistence, and parallel work
  • Authentication patterns including import-from-browser, persistent profiles, and OAuth/2FA flows
  • Device, viewport, proxy, and network controls for robust testing from the command line
  • Recording and debugging aids: screenshots, PDF capture, video recording, and diffing

Quick Start

Open a website, snapshot the page to generate refs, then interact with elements using those refs.

Frequently Asked Questions about agent-browser

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate web testing and browser interactions from the command line?

Automate web testing by using a CLI to drive Chrome or Chromium via the Chrome DevTools Protocol. This enables deterministic actions like opening pages, filling forms, clicking elements, taking screenshots, and scraping data.

How does browser automation handle element interactions for AI agents?

Browser automation for AI agents handles interactions by capturing a page snapshot to generate low-context refs. You then interact with elements using these @refs to perform clicks, fills, and navigation deterministically.

Can I manage authentication and session states when automating browser workflows?

Session management in browser automation supports named sessions, state persistence, and parallel work. Authentication patterns include importing states from an existing browser, using persistent profiles, and handling OAuth or 2FA flows.

Does this CLI browser automation tool support proxies and network controls for testing?

The CLI browser automation tool supports device, viewport, proxy, and network controls. This allows you to configure robust testing environments directly from the command line for various web app automation scenarios.

What debugging aids are available for web scraping and browser automation tasks?

Debugging aids for browser automation include screenshots, PDF capture, video recording, and visual diffing. These tools help verify web scraping, form filling, and automated testing workflows executed via the CLI.

What are the limitations of using CDP for browser automation in frames?

Using the Chrome DevTools Protocol for browser automation requires handling frames explicitly via snapshot refs. It necessitates a local or remote Chrome/Chromium instance and may require careful session management for complex parallel workflows.