Agent助手

Name: Agent助手
Author: 00OO666

AI 智能社区

作者仓库星标 4
作者更新于实时读取
作者仓库 agent-browser-skill

流狐档案事实层信息：领域、Agent、信任分、运行环境、权限与网络

领域: AI 智能
兼容 Agent: Claude Code
Cursor
Cline
Codex
Windsurf
Gemini CLI

+20
信任分: 88 / 100 · 社区维护
作者 / 版本 / 许可: @00OO666 · 未声明 license
Token 消耗评级: 较高消耗
接入复杂程度: 需简单配置
是否需要外部 API Key: 不需要
兼容的系统: macOS · Linux · Windows
底层运行要求: Node.js
文件与系统权限: 只读
允许写入 / 修改
Shell 执行
读取环境变量
网络行为: 允许外网请求
安装命令数: 26 条

档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。

需要注意： 未限定 allowed-tools，默认拥有全部工具权限。

输出预览 agent-browser-installer.preview

---
name: agent-browser-installer
description: | Automated installer that sets up the complete agent-browser environment with a single command.…
category: AI 智能
runtime: Node.js
---

# agent-browser-installer 输出预览

## PART A: 任务判断
- 适用问题：提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求：目标材料、限制条件、期望输出和验收方式。
- 证据边界：围绕“🎯 Overview / ⚡ Execution Rules / 📋 Installation Workflow”读取原文规则，不把推断写成作者承诺。

## PART B: 执行结果
- **01** 任务判断：确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理，并标出输入、限制和预期结果。
- **02** 执行计划：优先按“🎯 Overview / ⚡ Execution Rules / 📋 Installation Workflow”拆成步骤，说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果：给出可复制的命令、文件改动、检查清单或内容草稿，并说明如何继续迭代。
- **04** 风险边界：结合 读取文件、写入/修改文件、执行终端命令、读取环境变量、会按任务需要访问外部网络、通常不需要额外 API Key 给出执行前确认项。

## Running Rules
- 读取文件、写入/修改文件、执行终端命令、读取环境变量；会按任务需要访问外部网络；通常不需要额外 API Key。
- 先小样例验证，再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。

SKILL.md SKILL.md

---
name: agent-browser-installer
description: | Automated installer that sets up the complete agent-browser environment with a single command.…
category: AI 智能
source: 00OO666/agent-browser-skill
---

# agent-browser-installer

## 什么时候使用
- agent-browser-installer 是 AI 智能方向的技能，主要扩展 Agent 在调模型、改提示词、跑评测这类场景下的能力 适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付…
- 面向提示词、Agent 工作流、模型评估或自动化推理，优先处理能明确输入、步骤和验收标准的工作。

## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。

## 执行规则
- 围绕「🎯 Overview / ⚡ Execution Rules / 📋 Installation Workflow」组织步骤，不把推断写成作者事实。
- 读取文件、写入/修改文件、执行终端命令、读取环境变量；会按任务需要访问外部网络；通常不需要额外 API Key。
- 先跑小样例，确认结果可检查后再扩大任务范围。

## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown，不编造命令、平台或依赖。

证据边界与执行链路

作者原文负责流程事实；仓库文件负责来源和命令；流狐只补充适用场景、限制和质量判断。

skill "agent-browser-installer" {
  输入层     -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
  上下文层   -> 🎯 Overview / ⚡ Execution Rules / 📋 Installation Workflow
  规则层     -> SKILL.md 触发条件 / 执行顺序 / 输出格式
  运行层     -> Node.js | 读取文件、写入/修改文件、执行终端命令、读取环境变量 | 会按任务需要访问外部网络
  安全层     -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
  输出层     -> 可复制结果 + 检查清单 + 下一步迭代
}

解读按原文结构重写，命令、链接、术语均保留；右侧可核对作者原始 SKILL.md

先判断是否适合

核心用途：agent-browser-installer 是 AI 智能方向的技能，主要扩展 Agent 在调模型、改提示词、跑评测这类场景下的能力适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免…
适用任务：更适合 AI Agent、提示词、模型评估与自动化推理，尤其是输入、步骤、验收口径可以反复复用的场景。
不宜硬套：如果任务只有一次性问答、来源材料很少，或需要作者没有说明的命令和外部服务，应先降级为普通对话。

作者设计意图

触发依据：使用时机、上下文收集、输出格式和失败处理共同决定它是否适合当前任务。
来源边界：作者原文负责事实，仓库文件负责命令和文件证据；Fluxly 只补充用途、限制和使用判断。

作者的方法与取舍

执行主线：任务围绕「🎯 Overview / ⚡ Execution Rules / 📋 Installation Workflow」组织，把触发条件、必要上下文、执行规则和交付标准连成一条可复核的链路。
设计重点：重点不在重复教程，而在解释作者为什么规定这些步骤、哪些条件不可省略，以及结果偏离时应回到哪一层检查。
使用判断：先确认任务与原文边界一致，再决定是否启用；来源没有说明的命令、平台和依赖保持未知，不由 Fluxly 补写。

边界和复核

依赖条件：通常不需要额外 API Key，适合先用小任务验证触发是否稳定。
权限提醒：会涉及 read / write / shell-exec / env-read 权限，执行前要说明可读写范围、命令边界和回退方式。
质量标准：结果应包含明确产物、必要证据和下一步动作；如果只是泛泛解释，需要补充输入和验收条件后重跑。

Agent Browser Installer

🎯 Overview

Automated installer that sets up the complete agent-browser environment with a single command. Handles all dependencies, creates CLI wrappers, and verifies the installation.

⚡ Execution Rules

When this Skill is invoked, you MUST:

Execute immediately - Don't just say "I will install", actually run the commands
Use AskUserQuestion tool to confirm installation paths and preferences
Use Bash tool to execute all installation commands
Use Write tool to create configuration files
Show progress and results to the user at each step
Verify installation before completing

📋 Installation Workflow

Step 1: Pre-Installation Check

Check system requirements and existing installations:

# Check Node.js version
node --version

# Check npm version
npm --version

# Check if agent-browser is already installed
npm list -g agent-browser 2>/dev/null || echo "Not installed"

# Check available disk space
df -h ~

Requirements:

Node.js 18+ (required)
npm 8+ (required)
500MB free disk space (recommended)

Step 2: Ask User Preferences

Use AskUserQuestion to gather:

Question 1: Installation Location

Header: "Install Path"
Options:
- Global npm directory (Recommended) - npm install -g
- Custom directory - User specifies path
Description: Global installation makes agent-browser available everywhere

Question 2: Browser Download

Header: "Browser"
Options:
- Auto-download Chromium (Recommended) - Playwright will download
- Use existing Chrome - Connect to Chrome debugging port
- Manual setup - User will install browser later
Description: Auto-download ensures compatibility

Question 3: CLI Wrapper

Header: "CLI Wrapper"
Options:
- Create ab alias (Recommended) - Short command for quick access
- Use full command - agent-browser only
- Both - Create alias and keep full command
Description: Alias makes commands shorter and easier to type

Step 3: Install agent-browser

Execute installation based on user preferences:

# Global installation (recommended)
npm install -g agent-browser

# Or custom directory
cd /path/to/custom/dir && npm install agent-browser

Handle common errors:

Permission denied → Suggest sudo npm install -g agent-browser (Linux/Mac)
Network timeout → Retry with npm install -g agent-browser --registry=https://registry.npmmirror.com
Lock file error → Clean npm cache: npm cache clean --force

Step 4: Install Chromium Browser

Download Chromium using Playwright:

# Install Chromium
npx playwright install chromium

# Verify installation
npx playwright --version

Alternative: Manual download (if auto-install fails)

For Windows:

# Create browser directory
mkdir -p ~/playwright-browsers/chromium-1208

# Download browser
curl -L "https://cdn.playwright.dev/chrome-for-testing-public/145.0.7632.6/win64/chrome-win64.zip" -o ~/playwright-browsers/chrome.zip

# Extract
unzip ~/playwright-browsers/chrome.zip -d ~/playwright-browsers/chromium-1208

# Clean up
rm ~/playwright-browsers/chrome.zip

For Linux:

curl -L "https://cdn.playwright.dev/chrome-for-testing-public/145.0.7632.6/linux/chrome-linux64.zip" -o ~/playwright-browsers/chrome.zip
unzip ~/playwright-browsers/chrome.zip -d ~/playwright-browsers/chromium-1208
rm ~/playwright-browsers/chrome.zip

For macOS:

curl -L "https://cdn.playwright.dev/chrome-for-testing-public/145.0.7632.6/mac/chrome-mac64.zip" -o ~/playwright-browsers/chrome.zip
unzip ~/playwright-browsers/chrome.zip -d ~/playwright-browsers/chromium-1208
rm ~/playwright-browsers/chrome.zip

Step 5: Create CLI Wrapper (Optional)

If user chose to create ab alias:

For Linux/macOS:

# Add to ~/.bashrc or ~/.zshrc
echo 'alias ab="agent-browser"' >> ~/.bashrc
source ~/.bashrc

For Windows (Git Bash):

# Add to ~/.bashrc
echo 'alias ab="agent-browser"' >> ~/.bashrc
source ~/.bashrc

For Windows (PowerShell):

# Add to PowerShell profile
Add-Content $PROFILE 'Set-Alias -Name ab -Value agent-browser'
. $PROFILE

Step 6: Create agent-browser Skill

Create the skill directory and SKILL.md:

# Create skill directory
mkdir -p ~/.claude/skills/agent-browser

# Skill will be created using Write tool (see below)

SKILL.md content (use Write tool):

---
name: agent-browser
description: |
  Browser automation CLI tool with session persistence and file upload.
  Triggers when user needs browser automation, web scraping, or page interaction.
  Core functions: (1) Navigate pages (2) Interact with elements (3) Take screenshots (4) Upload files (5) Persist sessions
---

# Agent Browser - Browser Automation Skill

## 🎯 Overview

AI-native browser automation tool built on Playwright. Uses ref-based element selection for 93% less context consumption.

## ⚡ Execution Rules

**When this Skill is invoked, you MUST:**

1. **Execute immediately** - Don't just describe, run the commands
2. **Use Bash tool** to execute agent-browser commands
3. **Show results** to the user
4. **Close browser** when done to avoid session leaks

## 📋 Command Reference

### Navigation
- `agent-browser open <url>` - Navigate to URL
- `agent-browser back` - Go back
- `agent-browser forward` - Go forward
- `agent-browser reload` - Reload page

### Element Interaction
- `agent-browser snapshot -i` - Get interactive elements with refs
- `agent-browser click @ref` - Click element
- `agent-browser fill @ref <text>` - Fill input field
- `agent-browser type @ref <text>` - Type into element
- `agent-browser hover @ref` - Hover over element

### Information Retrieval
- `agent-browser get text <selector>` - Get text content
- `agent-browser get title` - Get page title
- `agent-browser get url` - Get current URL
- `agent-browser screenshot [path]` - Take screenshot

### File Operations
- `agent-browser upload @ref <filepath>` - Upload file

### Session Persistence
- `agent-browser --profile <path> open <url>` - Use persistent profile
- Profile saves cookies, localStorage, and login states

### Browser Control
- `agent-browser close` - Close browser
- `agent-browser connect [endpoint]` - Connect to Chrome debugging port

## 💡 Usage Examples

### Example 1: Simple Navigation
User: "Open Google and take a screenshot"

**Execute immediately:**
```bash
agent-browser open https://google.com
agent-browser screenshot ~/Desktop/google.png
agent-browser close

Example 2: Session Persistence

User: "Login to GitHub and save the session"

Execute immediately:

# First time: login and save
agent-browser --profile ~/.agent-browser/github --headed open https://github.com/login
# (User logs in manually)

# Next time: automatically logged in
agent-browser --profile ~/.agent-browser/github open https://github.com/settings

Example 3: File Upload

User: "Upload my avatar to GitHub profile"

Execute immediately:

agent-browser --profile ~/.agent-browser/github open https://github.com/settings/profile
agent-browser snapshot -i  # Find file input ref (e.g., @e68)
agent-browser upload @e68 "/path/to/avatar.jpg"
agent-browser click @save-button-ref
agent-browser close

Example 4: Form Filling

User: "Fill out the contact form on example.com"

Execute immediately:

agent-browser open https://example.com/contact
agent-browser snapshot -i  # Get form element refs
agent-browser fill @e1 "John Doe"
agent-browser fill @e2 "john@example.com"
agent-browser fill @e3 "Hello, this is a test message"
agent-browser click @e4  # Submit button
agent-browser close

🔧 Advanced Features

Batch Operations

Execute multiple commands at once:

agent-browser batch "click @e1" "fill @e2 value" "click @e3"

Connect to Existing Chrome

Use your existing Chrome with all logins:

# Start Chrome with debugging
chrome --remote-debugging-port=9222

# Connect agent-browser
agent-browser connect http://127.0.0.1:9222
agent-browser open https://gmail.com  # Already logged in!

⚠️ Important Notes

Always close browser - Use agent-browser close to avoid session leaks
Use profiles for persistence - --profile flag saves login states
Get refs before interaction - Run snapshot -i before click or fill
Absolute paths for files - Use full paths for screenshots and uploads
Wait for page load - Add agent-browser wait 2000 if needed

🎯 Trigger Scenarios

Use this skill when user says:

"Open a webpage..."
"Browse..."
"Take a screenshot..."
"Scrape web content..."
"Click on..."
"Fill out a form..."
"Upload a file..."
"Automate browser..."

Version: v1.0.0 Status: ✅ Ready to use


### Step 7: Verification

Run comprehensive tests to verify installation:

```bash
# Test 1: Check agent-browser is installed
which agent-browser || echo "❌ agent-browser not found in PATH"

# Test 2: Check version
agent-browser --version || echo "❌ Failed to get version"

# Test 3: Test basic navigation
agent-browser open https://example.com
agent-browser get title
agent-browser close

# Test 4: Test screenshot
agent-browser open https://example.com
agent-browser screenshot /tmp/test-screenshot.png
ls -lh /tmp/test-screenshot.png
agent-browser close

# Test 5: Test snapshot
agent-browser open https://example.com
agent-browser snapshot -i
agent-browser close

# Test 6: Check Chromium installation
ls -la ~/playwright-browsers/chromium-*/chrome* || echo "⚠️ Chromium not found"

Verification Checklist:

✅ agent-browser command available
✅ Version command works
✅ Can open webpages
✅ Can take screenshots
✅ Can get page snapshots
✅ Chromium browser installed
✅ Skill created in ~/.claude/skills/

Step 8: Generate Installation Report

Create a summary report:

# Agent Browser Installation Report

## ✅ Installation Summary

**Date**: {timestamp}
**Status**: SUCCESS / FAILED

## 📦 Installed Components

- ✅ agent-browser CLI: v{version}
- ✅ Chromium Browser: {browser_version}
- ✅ CLI Wrapper: {ab/full/both}
- ✅ Claude Skill: ~/.claude/skills/agent-browser/

## 🔧 Configuration

- Installation Path: {path}
- Browser Path: {browser_path}
- Profile Directory: ~/.agent-browser/

## 🚀 Quick Start

### Basic Usage
```bash
agent-browser open https://example.com
agent-browser screenshot ~/Desktop/screenshot.png
agent-browser close

With Session Persistence

agent-browser --profile ~/.agent-browser/myprofile open https://github.com

📚 Next Steps

Try the examples in the skill documentation
Create profiles for different accounts
Explore advanced features (batch operations, file upload)

🆘 Troubleshooting

If you encounter issues:

Check Node.js version: node --version (need 18+)
Verify installation: agent-browser --version
Check browser: ls ~/playwright-browsers/
See documentation: ~/.claude/skills/agent-browser/SKILL.md

Installation completed successfully! 🎉


## 💡 Usage Examples

### Example 1: Full Installation (Recommended)

User: "Install agent-browser"

**Execute immediately:**

1. Ask user preferences using AskUserQuestion
2. Run pre-installation checks
3. Install agent-browser globally
4. Download Chromium
5. Create `ab` alias
6. Create skill
7. Verify installation
8. Show installation report

### Example 2: Reinstall/Repair

User: "Reinstall agent-browser, it's not working"

**Execute immediately:**

1. Uninstall existing: `npm uninstall -g agent-browser`
2. Clean cache: `npm cache clean --force`
3. Remove browser: `rm -rf ~/playwright-browsers/`
4. Follow full installation workflow
5. Verify installation

### Example 3: Update Installation

User: "Update agent-browser to latest version"

**Execute immediately:**

```bash
# Update agent-browser
npm update -g agent-browser

# Update Chromium
npx playwright install chromium

# Verify versions
agent-browser --version
npx playwright --version

🔧 Troubleshooting Guide

Issue 1: Permission Denied (Linux/Mac)

Error: EACCES: permission denied

Solution:

# Option 1: Use sudo (not recommended)
sudo npm install -g agent-browser

# Option 2: Fix npm permissions (recommended)
mkdir ~/.npm-global
npm config set prefix '~/.npm-global'
echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.bashrc
source ~/.bashrc
npm install -g agent-browser

Issue 2: Network Timeout

Error: network timeout or ETIMEDOUT

Solution:

# Use mirror registry
npm install -g agent-browser --registry=https://registry.npmmirror.com

# Or increase timeout
npm install -g agent-browser --fetch-timeout=60000

Issue 3: Chromium Download Failed

Error: Failed to download Chromium

Solution: Use manual download (see Step 4 alternative methods)

Issue 4: Command Not Found

Error: agent-browser: command not found

Solution:

# Check npm global bin path
npm config get prefix

# Add to PATH
export PATH="$(npm config get prefix)/bin:$PATH"
echo 'export PATH="$(npm config get prefix)/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

Issue 5: Playwright Lock File Error

Error: Unable to update lock within the stale threshold

Solution:

# Remove lock files
rm -rf ~/playwright-browsers/__dirlock

# Retry installation
npx playwright install chromium

📚 Post-Installation

Create Your First Profile

# Create profile directory
mkdir -p ~/.agent-browser/github

# Use profile for persistent sessions
agent-browser --profile ~/.agent-browser/github --headed open https://github.com/login

Set Up Environment Variables (Optional)

# Add to ~/.bashrc or ~/.zshrc
export AGENT_BROWSER_PROFILE=~/.agent-browser/default
export AGENT_BROWSER_HEADLESS=true

Test Advanced Features

# Test batch operations
agent-browser open https://example.com
agent-browser batch "snapshot -i" "screenshot /tmp/test.png" "close"

# Test file upload
agent-browser open https://example.com/upload
agent-browser snapshot -i
agent-browser upload @e1 "/path/to/file.pdf"

⚠️ Important Notes

Node.js 18+ required - Check with node --version
500MB disk space - For Chromium browser
Internet connection - For downloading packages
Administrator rights - May be needed for global installation
Firewall settings - Allow Node.js and Chromium

🎯 Success Criteria

Installation is successful when:

✅ agent-browser --version returns version number
✅ agent-browser open https://example.com works
✅ Screenshots can be taken
✅ Skill appears in ~/.claude/skills/agent-browser/
✅ All verification tests pass

Version: v1.0.0 Last Updated: 2026-01-27 Status: ✅ Ready to use

安装命令

一键复制安装选择包管理器,点击右侧复制,粘贴到终端回车即可

npx skills add 00OO666/agent-browser-skill/agent-browser-installer -y

查看通用安装教程

查看 Skill 原文

兼容工具

Claude Code
Cursor
Cline
Codex
Windsurf
Gemini
OpenCode
Roo Code
AAide
+Augment
GitHub Copilot
OpenClaw
KKiro
Junie
WWarp
Goose
Trae
Kilo Code
OpenHands
Amp
Zencoder
CContinue
AIAider
CodeBuddy
CodeGeeX
Replit Agent

安全审计

88 / 100

社区

未声明 license/许可。
未声明 version。
未限定 allowed-tools，默认拥有全部工具权限。

元信息

来源: github
00OO666/agent-browser-skill
English

@00OO666 的其他技能

A Agent助手 88

分类与标签

分类 AI 智能

先判断是否适合

作者设计意图

作者的方法与取舍

边界和复核

Agent Browser Installer

🎯 Overview

⚡ Execution Rules

📋 Installation Workflow

Step 1: Pre-Installation Check

Step 2: Ask User Preferences

Step 3: Install agent-browser

Step 4: Install Chromium Browser

Step 5: Create CLI Wrapper (Optional)

Step 6: Create agent-browser Skill

Example 2: Session Persistence

Example 3: File Upload

Example 4: Form Filling

🔧 Advanced Features

Batch Operations

Connect to Existing Chrome

⚠️ Important Notes

🎯 Trigger Scenarios

Step 8: Generate Installation Report

With Session Persistence

📚 Next Steps

🆘 Troubleshooting

🔧 Troubleshooting Guide

Issue 1: Permission Denied (Linux/Mac)

Issue 2: Network Timeout

Issue 3: Chromium Download Failed

Issue 4: Command Not Found

Issue 5: Playwright Lock File Error

📚 Post-Installation

Create Your First Profile

Set Up Environment Variables (Optional)

Test Advanced Features

⚠️ Important Notes

🎯 Success Criteria

讨论