agent-browser

by @ZenoWangzy in Development

# Install this skill:

npx skills add ZenoWangzy/agent-browser-skill

Or install specific skill: npx add-skill https://github.com/ZenoWangzy/agent-browser-skill

# Description

Browser automation CLI for AI agents with Rust + Node.js architecture. Use when ANY browser/web automation is needed: (1) Web scraping & data extraction, (2) Form filling & submission, (3) Screenshot & PDF generation, (4) Testing web applications (E2E, forms, navigation), (5) Cookie/storage management, (6) Multi-tab/window operations, (7) Element interaction (click, type, hover, drag), (8) Waiting & state verification, (9) Page navigation & history, (10) Network request mocking, (11) Browser automation coding (understanding architecture, API patterns, test examples, navigating codebase). Ideal for contributors, developers, and ANYONE automating browser interactions.

# SKILL.md

name: agent-browser
description: Browser automation CLI for AI agents with Rust + Node.js architecture. Use when ANY browser/web automation is needed: (1) Web scraping & data extraction, (2) Form filling & submission, (3) Screenshot & PDF generation, (4) Testing web applications (E2E, forms, navigation), (5) Cookie/storage management, (6) Multi-tab/window operations, (7) Element interaction (click, type, hover, drag), (8) Waiting & state verification, (9) Page navigation & history, (10) Network request mocking, (11) Browser automation coding (understanding architecture, API patterns, test examples, navigating codebase). Ideal for contributors, developers, and ANYONE automating browser interactions.

agent-browser Codebase

Fast headless browser automation CLI for AI agents. Rust CLI with Node.js fallback, built on Playwright.

Architecture Overview

┌─────────────────────┐
│   Rust CLI          │  ← Fast native binary (commands.rs, main.rs)
│   (Fast parsing)    │
└──────────┬──────────┘
           │
           ↓
┌─────────────────────┐
│   Node.js Daemon    │  ← Playwright browser management
│   (Browser control) │
└─────────────────────┘

Languages: TypeScript 72.7% | Rust 21.2% | JavaScript 6.1%

When to Use This Skill

Browser Automation Usage (日常使用)

场景	参考文档
网页爬取/数据提取	`references/examples/workflows.md` - Web Scraping Workflow
表单填写/提交	`references/examples/workflows.md` - Form Testing Workflow
截图/PDF生成	`references/examples/quick-start.md` - Screenshot & PDF
网页测试 (E2E)	`references/examples/workflows.md` - E2E Testing Workflow
Cookie/Storage管理	`references/examples/quick-start.md` - Cookie & Storage
多标签页操作	`references/examples/workflows.md` - Multi-Tab Workflow
元素交互	`references/examples/quick-start.md` - Common Commands
等待/状态验证	`references/examples/workflows.md` - Wait Strategies
页面导航	`references/examples/quick-start.md` - Navigation
网络请求模拟	`references/examples/workflows.md` - Network Mocking

Codebase Analysis (代码分析)

任务	参考文档
API 实现	`references/api_reference/actions.ts` - 100+ 命令处理器
CLI 命令解析	`references/api_reference/commands.rs` - Rust 解析器
浏览器管理	`references/api_reference/browser.ts`
测试模式	`references/api_reference/*.test.ts` - 测试示例
配置分析	`references/config_patterns/` - 15 个配置文件

Quick Links

All Commands: references/api_reference/actions.md (click, fill, screenshot, etc.)
Design Patterns: references/patterns/detected_patterns.json (Factory Pattern)
Dependencies: references/dependencies/dependency_graph.mmd (44 nodes, 0 cycles)
Usage Examples: references/examples/quick-start.md - Real workflows from README

cli/src/          → Rust CLI (commands, flags, output)
src/              → TypeScript daemon (browser, actions, protocol)
docs/src/app/     → Next.js documentation site
test/             → Test suites

Generated by Skill Seeker | Enhanced with skill-creator best practices

# README.md

agent-browser Skill

Fast headless browser automation CLI for AI agents - Claude Code Skill

🚀 Quick Install

claude add ZenoWangzy/agent-browser-skill

Or download manually:

# Clone this repository
git clone https://github.com/ZenoWangzy/agent-browser-skill.git ~/.claude/skills/agent-browser

📖 What is this?

This is a Claude Code Skill for the agent-browser project - a fast headless browser automation CLI built with Rust and Node.js (Playwright).

Architecture

┌─────────────────────┐
│   Rust CLI          │  ← Fast native binary (commands.rs, main.rs)
│   (Fast parsing)    │
└──────────┬──────────┘
           │
           ↓
┌─────────────────────┐
│   Node.js Daemon    │  ← Playwright browser management
│   (Browser control) │
└─────────────────────┘

Languages: TypeScript 72.7% | Rust 21.2% | JavaScript 6.1%

🎯 Use Cases

Browser Automation

Scenario	Reference
🕷️ Web Scraping	`references/examples/workflows.md`
📝 Form Filling	`references/examples/workflows.md`
📸 Screenshots/PDF	`references/examples/quick-start.md`
🧪 E2E Testing	`references/examples/workflows.md`
🍪 Cookie/Storage	`references/examples/quick-start.md`
📑 Multi-tab Operations	`references/examples/workflows.md`
🖱️ Element Interaction	`references/examples/quick-start.md`
⏳ Wait Strategies	`references/examples/workflows.md`
🌐 Page Navigation	`references/examples/quick-start.md`
🌐 Network Mocking	`references/examples/workflows.md`

Codebase Analysis

Task	Reference
API Implementation	`references/api_reference/actions.md`
CLI Command Parsing	`references/api_reference/commands.md`
Browser Management	`references/api_reference/browser.md`
Test Patterns	`references/api_reference/*.test.md`
Configuration	`references/config_patterns/`

📚 Documentation Structure

.
├── SKILL.md                    # Main skill file for Claude Code
├── README.md                   # This file
└── references/
    ├── examples/
    │   ├── quick-start.md      # Quick start guide with examples
    │   ├── workflows.md        # Common workflows
    │   └── integration.md      # Integration guide
    ├── api_reference/          # API documentation (25+ files)
    ├── config_patterns/        # Configuration analysis
    ├── dependencies/           # Dependency graphs
    └── patterns/               # Design patterns

🔧 Quick Links

All Commands: references/api_reference/actions.md (click, fill, screenshot, etc.)
Design Patterns: references/patterns/detected_patterns.json
Dependencies: references/dependencies/dependency_graph.mmd
Usage Examples: references/examples/quick-start.md

💡 Usage Example

After installing this skill, Claude Code will automatically use it when you need browser automation tasks. Example:

You: Take a screenshot of https://example.com

Claude: I'll use agent-browser to take a screenshot...
[Uses the skill to navigate and capture the page]

🤝 Contributing

This skill is generated from the agent-browser codebase. To update:

Update the agent-browser codebase
Re-run the skill generation
Submit a PR to this repository

📄 License

Same as the agent-browser project.

Made with ❤️ for the Claude Code community

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

⚡ Amp 🚀 Antigravity 🤖 Claude Code 🦀 Clawdbot 📝 Codex ▶️ Cursor 🤖 Droid 💎 Gemini CLI 🐙 GitHub Copilot 🪿 Goose 📊 Kilo Code 🔧 Kiro CLI 💻 OpenCode 🦘 Roo Code 🌲 Trae 🏄 Windsurf

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.