Token导航 LogoToken导航TokenDH.com
中转站排行榜
研究检索操作浏览器github未标认证来源可访问许可证需确认审计异常

agent-browser（Agent 浏览器自动化）

Agent Skill

agent-browser 用于处理浏览器自动化、网页检查和页面信息提取，适合在 Codex、Claude、Cursor、Gemini CLI 中需要让 Agent 打开页面、读取网页或验证前端流程时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态，以及是否会触发联网、命令执行或文件读写。

总安装

346

周安装

14

GitHub Stars

5

下载量

109
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息，不托管安装包，也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手，先确认命令和权限，再让它执行。

请帮我安装这个 Agent Skill：agent-browser（Agent 浏览器自动化）
来源仓库：https://github.com/manaflow-ai/cloudrouter
仓库路径：skills/agent-browser
安装命令：
npx skills add https://github.com/manaflow-ai/cloudrouter --skill agent-browser
安装前请先检查当前环境是否支持对应 CLI，并向我确认将要执行的命令、安装目录、联网范围和文件读写权限；确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill；本站只展示命令，不托管安装包，也不自动执行。

skills.shnpx skills
npx skills add https://github.com/manaflow-ai/cloudrouter --skill agent-browser

简介

用于浏览器自动化和网页信息提取。agent-browser 属于研究检索类 Skill，可作为该场景下的辅助能力补充。

  • 适合页面打开、内容读取或前端流程验证。适用宿主包括 Codex、Claude、Cursor、Gemini CLI，接入前应确认版本、权限和运行环境要求。
  • 通过 GitHub 安装，需确认权限和网络访问能力。
  • 可能触发实际浏览器操作或 DOM 解析。
  • 建议结合原始 README 了解安全边界。

SKILL.md

Browser Automation via cloudrouter browser

Automate browser interactions in cloud sandboxes. cloudrouter browser wraps agent-browser and runs commands inside the sandbox via SSH.

Prerequisite: You need a running cloudrouter sandbox. Get the sandbox ID from cloudrouter ls or create one with cloudrouter start..

Core Workflow

Always follow this pattern: Open -> Snapshot -> Interact -> Re-snapshot

cloudrouter browser open <id> "https://example.com"
cloudrouter browser snapshot -i <id>         # -i = interactive elements only
cloudrouter browser click <id> @e5           # Click element ref
cloudrouter browser fill <id> @e3 "text"     # Fill input
cloudrouter browser snapshot -i <id>         # Re-snapshot after change

Essential Commands

Navigation

cloudrouter browser open <id> <url>          # Navigate to URL
cloudrouter browser back <id>                # Go back
cloudrouter browser forward <id>             # Go forward
cloudrouter browser reload <id>              # Refresh page
cloudrouter browser url <id>                 # Get current URL
cloudrouter browser title <id>               # Get page title

Inspection

cloudrouter browser snapshot -i <id>         # Interactive elements only (RECOMMENDED)
cloudrouter browser snapshot <id>            # Full accessibility tree
cloudrouter browser snapshot -c <id>         # Compact output
cloudrouter browser screenshot <id>          # Screenshot (base64 to stdout)
cloudrouter browser screenshot <id> out.png  # Save screenshot to file
cloudrouter browser screenshot --full <id>   # Full page screenshot
cloudrouter browser eval <id> "document.title"  # Run JavaScript

Interaction

cloudrouter browser click <id> @e1           # Click element
cloudrouter browser dblclick <id> @e1        # Double-click
cloudrouter browser fill <id> @e2 "text"     # Clear input and type text
cloudrouter browser type <id> @e2 "text"     # Type without clearing (appends)
cloudrouter browser press <id> Enter         # Press key (Enter, Tab, Escape, etc.)
cloudrouter browser hover <id> @e5           # Hover over element
cloudrouter browser focus <id> @e3           # Focus element
cloudrouter browser scroll <id> down         # Scroll down (default pixels)
cloudrouter browser scroll <id> down 500     # Scroll down 500px
cloudrouter browser scroll <id> up           # Scroll up
cloudrouter browser scrollintoview <id> "#element"  # Scroll into view (CSS selector only, NOT @e refs)
cloudrouter browser select <id> @e7 "value"  # Select dropdown option
cloudrouter browser check <id> @e8           # Check checkbox
cloudrouter browser uncheck <id> @e9         # Uncheck checkbox
cloudrouter browser upload <id> @e10 /path   # Upload file
cloudrouter browser drag <id> @e1 @e2        # Drag and drop
cloudrouter browser wait <id> @e3            # Wait for element to appear
cloudrouter browser wait <id> 2000           # Wait milliseconds

Get Information

cloudrouter browser get-text <id> @e1        # Get element text
cloudrouter browser get-value <id> @e2       # Get input value
cloudrouter browser get-attr <id> @e3 href   # Get attribute
cloudrouter browser get-html <id> @e4        # Get innerHTML
cloudrouter browser get-count <id> ".item"   # Count matching elements
cloudrouter browser is-visible <id> @e1      # Check visibility
cloudrouter browser is-enabled <id> @e1      # Check if enabled
cloudrouter browser is-checked <id> @e1      # Check if checked

Semantic Locators (alternative to refs)

When refs are unreliable on dynamic pages:

cloudrouter browser find <id> text "Sign In" click                 # By visible text
cloudrouter browser find <id> label "Email" fill "user@test.com"   # By label
cloudrouter browser find <id> placeholder "Search" type "query"    # By placeholder
cloudrouter browser find <id> testid "submit-btn" click            # By data-testid
Note: find <id> role button click finds the FIRST button on the page — it cannot filter by name. Use find <id> text "Button Name" click to target a specific button. There is no --name flag.

JavaScript & Debugging

cloudrouter browser eval <id> "document.title"          # Evaluate JS
cloudrouter browser console <id>                        # View console output
cloudrouter browser errors <id>                         # View JS errors

Tabs & Frames

cloudrouter browser tab-list <id>                       # List tabs
cloudrouter browser tab-new <id> "https://..."          # New tab
cloudrouter browser tab-switch <id> 2                   # Switch tab
cloudrouter browser tab-close <id>                      # Close tab
cloudrouter browser frame <id> "#iframe"                # Switch to iframe
cloudrouter browser frame <id> main                     # Back to main

Cookies & Storage

cloudrouter browser cookies <id>                        # List cookies
cloudrouter browser cookies-set <id> name value         # Set cookie
cloudrouter browser cookies-clear <id>                  # Clear cookies
cloudrouter browser storage-local <id>                  # Get localStorage
cloudrouter browser storage-local-set <id> key value    # Set localStorage
cloudrouter browser storage-local-clear <id>            # Clear localStorage

State Management

cloudrouter browser state-save <id> /tmp/auth.json      # Save cookies + storage
cloudrouter browser state-load <id> /tmp/auth.json      # Restore state

Browser Settings

cloudrouter browser set-viewport <id> 1920 1080         # Set viewport
cloudrouter browser set-device <id> "iPhone 14"         # Emulate device
cloudrouter browser set-geo <id> 37.77 -122.42          # Set geolocation
cloudrouter browser set-offline <id> on                 # Toggle offline
cloudrouter browser set-media <id> dark                 # Color scheme

Network Interception

cloudrouter browser network-route <id> "**/api/*"       # Intercept requests
cloudrouter browser network-route <id> "**/ads/*" --abort  # Block requests
cloudrouter browser network-unroute <id>                # Remove routes
cloudrouter browser network-requests <id>               # List requests

Dialogs

cloudrouter browser dialog-accept <id>                  # Accept alert/confirm
cloudrouter browser dialog-accept <id> "answer"         # Accept prompt with text
cloudrouter browser dialog-dismiss <id>                 # Dismiss dialog

Element Selectors

  • Element refs from snapshot: @e1, @e2, @e3... (preferred)
  • CSS selectors: #id, .class, button[type="submit"]

Snapshot output shows [ref=e1] — use as @e1 in commands.

Common Patterns

Login Flow

cloudrouter browser open <id> "https://app.example.com/login"
cloudrouter browser snapshot -i <id>
# → @e1 [input] Email, @e2 [input] Password, @e3 [button] Sign In
cloudrouter browser fill <id> @e1 "user@example.com"
cloudrouter browser fill <id> @e2 "password123"
cloudrouter browser click <id> @e3
cloudrouter browser wait <id> 2000
cloudrouter browser snapshot -i <id>         # Verify login success
cloudrouter browser screenshot <id> /tmp/result.png

Form Submission

cloudrouter browser open <id> "https://example.com/contact"
cloudrouter browser snapshot -i <id>
cloudrouter browser fill <id> @e1 "John Doe"
cloudrouter browser fill <id> @e2 "john@email.com"
cloudrouter browser fill <id> @e3 "Hello world"
cloudrouter browser click <id> @e4           # Submit
cloudrouter browser wait <id> 2000
cloudrouter browser snapshot -i <id>         # Verify submission

Data Extraction

cloudrouter browser open <id> "https://example.com/products"
cloudrouter browser snapshot <id>            # Full tree for structure
cloudrouter browser get-text <id> @e5        # Extract specific text
cloudrouter browser eval <id> "JSON.stringify([...document.querySelectorAll('.product')].map(p => p.textContent))"

Multi-page Navigation

cloudrouter browser open <id> "https://example.com"
cloudrouter browser snapshot -i <id>
cloudrouter browser click <id> @e3           # Click a link
cloudrouter browser wait <id> 2000           # Wait for page load
cloudrouter browser snapshot -i <id>         # ALWAYS re-snapshot after navigation

Auth State Persistence

# Login once and save state
cloudrouter browser open <id> "https://app.example.com/login"
cloudrouter browser snapshot -i <id>
cloudrouter browser fill <id> @e1 "user@example.com"
cloudrouter browser fill <id> @e2 "password"
cloudrouter browser click <id> @e3
cloudrouter browser wait <id> 2000
cloudrouter browser state-save <id> /tmp/auth.json

# Restore in future sessions
cloudrouter browser state-load <id> /tmp/auth.json
cloudrouter browser open <id> "https://app.example.com/dashboard"

Critical Rules

  1. Flags go BEFORE the sandbox ID. cloudrouter browser snapshot -i <id> works. cloudrouter browser snapshot <id> -i silently returns empty/wrong results.
  2. ALWAYS re-snapshot after navigation or clicks. Page content changes, refs become stale.
  3. Use -i flag for snapshots — interactive elements only, much more efficient.
  4. Don't mix snapshot modes. Full snapshot and snapshot -i assign DIFFERENT ref numbers. Stick to one mode (use -i).
  5. Use fill not type for form fields. fill clears first; type appends.
  6. Refs are temporary. They reset after each snapshot. Always use fresh refs.
  7. Verify before interacting. Check snapshot output to confirm you have the right element.
  8. Handle loading states. If elements are missing, wait and re-snapshot.

Troubleshooting

IssueSolution
Element not found / wrong elementRe-snapshot with -i, refs are stale
snapshot <id> -i returns emptyPut flags BEFORE id: snapshot -i <id>
Click doesn't workTry hover first, then click
Page not loadingCheck URL with cloudrouter browser url <id>
Browser not readyWait a few seconds after sandbox creation, retry
Refs differ between snapshotsDon't mix full and -i snapshots
Form field has old textUse fill (clears first) instead of type
find... role button click clicks wrong buttonUse find... text "Button Name" click instead
find... --name "X" failsThere is no --name flag — use text locator
npm install EACCES errorRun cloudrouter ssh <id> "sudo chown -R 1000:1000 /home/user/.npm" first
scrollintoview @e1 fails with "Unsupported token"scrollintoview and highlight only accept CSS selectors ("#id", ".class"), NOT @e refs
pkill -f kills SSH session (exit 143/255)pkill -f pattern may match the SSH session. Just run another command to recover
pdf saves to remote, not localFile saves inside sandbox (e.g. /tmp/page.pdf). Use cloudrouter download to retrieve
storage-local <id> key shows "Done" not valueUse eval <id> "localStorage.getItem('key')" to get a specific localStorage value
Stale ref error: "Action timed out"Always re-snapshot after clicks/form submits that change the DOM

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明，方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用；具体调用方式请以来源页面和 README 为准。

平台分布

Codex

32.8%
按下载量换算36

Claude

31.05%
按下载量换算34

Cursor

20.51%
按下载量换算22

Gemini CLI

10.61%
按下载量换算12

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力，使用时可能读取或操作网页内容，需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息，不托管安装包，不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果，不能写成本站确认安全。当前只有一个来源，正式发布前建议补源仓库或其他目录站核验。

来源信息

GitHub 仓库

github.com

/manaflow-ai/cloudrouter

SKILL.md 来源

github.com

/manaflow-ai/cloudrouter/tree/main/skills/agent-browser

skills-sh

skills.sh

/manaflow-ai/cloudrouter/agent-browser

继续浏览同类 Skills

更多研究检索 Skills（73,675）返回 Skills 目录

上一个 Skill

agent-browser（Agent 浏览器自动化）

agent-browser 用于处理浏览器自动化、网页检查和页面信息提取，适合在 Codex、Claude、Cursor、Gemini CLI 中需要让 Agent 打开页面、读取网页或验证前端流程时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态，以及是否会触发联网、命令执行或文件读写。

下一个 Skill

agent-browser（Agent 浏览器自动化）

agent-browser 用于处理浏览器自动化、网页检查和页面信息提取，适合在 Codex、Claude、Cursor、Gemini CLI 中需要让 Agent 打开页面、读取网页或验证前端流程时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态，以及是否会触发联网、命令执行或文件读写。