agent-browser

Automate Chrome browser navigation, clicks, and data extraction via CLI.

641|104|Updated Mar 5, 2026
One-click install
npx skills add https://github.com/loongclaw-ai/loongclaw --skill agent-browser-loongclaw-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/loongclaw-ai/loongclaw/tree/main/skills/agent-browser
Command: npx skills add https://github.com/loongclaw-ai/loongclaw --skill agent-browser-loongclaw-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It removes the manual coordination of browser navigation, DOM interrogation, and interaction tasks by routing everything through the agent-browser CLI so agents can reliably perform web automation without juggling low-level CDP details.

Core Features & Use Cases

  • Deterministic workflows: Navigate pages, capture interactive refs, and chain fills/clicks while the daemon maintains persistent Chrome sessions.
  • Authentication and state management: Import browser auth, manage persistent profiles, save/load state files, and use the auth vault to keep credentials out of LLM output while handling OAuth, 2FA, and session expiry.
  • Evidence collection & observability: Capture screenshots, PDFs, annotated views, recordings, network har files, and performance traces, then compare snapshots to verify UI changes or regressions for testing and documentation.

Quick Start

Tell the agent to open the target URL with agent-browser, snapshot to gather refs, and then use documented clicks and fills to complete the workflow.

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 navigation and form filling for AI agents?

Automate web navigation and form filling for AI agents by routing interactions through the agent-browser CLI, which manages deterministic workflows like clicks, data extraction, and DOM interactions in persistent Chrome sessions.

How does browser session management handle authentication and OAuth flows?

Browser session management handles authentication and OAuth flows by importing browser auth, managing persistent profiles, and utilizing an auth vault to keep credentials out of LLM output while handling 2FA and session expiry.

Can I capture screenshots and network HAR files for UI regression testing?

Yes, you can capture screenshots and network HAR files for UI regression testing by collecting evidence through the agent-browser CLI, which supports annotated views, PDFs, recordings, and performance traces to verify UI changes.

What is the best way to extract data from web pages without juggling low-level CDP details?

The best way to extract data from web pages without low-level CDP details is using the agent-browser CLI to snapshot pages, gather interactive refs, and chain documented fills and clicks for reliable DOM interactions.

Does agent-browser work with Chrome and Chromium-based environments for web automation?

Yes, agent-browser works with Chrome and Chromium-based environments for web automation, maintaining persistent sessions through a daemon that reliably handles navigation, waiting, and DOM interactions.

Why does web automation fail when agents lose session state during navigation?

Web automation fails when agents lose session state because low-level browser coordination drops context; agent-browser solves this by using state files and persistent profiles to maintain reliable navigation across workflows.