Agent Browser Automation

reason-machines/trending-skills/skills/agent-browser-automation

作者 reason-machines2384a003145a無授權條款83 個星標收錄於 2026年10月8日更新於 2026年10月8日儲存庫3 個月前更新

Headless browser automation CLI for AI agents using native Rust binary with Chrome DevTools Protocol

僅含說明Software Development
AI 產生的概覽

介紹一款無頭瀏覽器自動化命令列工具，讓代理透過 Chrome DevTools 協定瀏覽、操作並擷取網頁。

功能
此技能提供 agent-browser 的使用說明，這是一套以 Rust 撰寫、透過 Chrome DevTools 協定驅動的無頭瀏覽器自動化命令列工具。內容涵蓋安裝、頁面導覽、附 @ref 元素識別碼的無障礙快照、點擊、表單填寫、螢幕擷取、匯出 PDF、網路攔截、Cookie 與儲存空間、分頁與 iframe，以及批次命令執行。也記錄了除錯、追蹤、視覺差異比對與工作階段狀態的儲存與載入。產出的是命令使用指引，而非指令碼。
適用情境
當代理需要操作真實瀏覽器時使用：擷取頁面、填寫表單、登入並重複使用工作階段、螢幕擷取或產生 PDF、模擬網路回應，或執行多步驟瀏覽器流程。適合以快照驅動的互動迴圈，此時無障礙參照比 CSS 選擇器更可靠。
執行需求
需要安裝 agent-browser 命令列工具（npm 全域、Homebrew 或 Cargo），並透過 agent-browser install 下載 Chrome for Testing；Linux 上可能需加上 --with-deps 安裝系統函式庫。安裝與瀏覽皆需網路存取。此技能不附指令碼，僅提供說明。

agent-browser

Skill by ara.so — Daily 2026 Skills collection.

agent-browser is a headless browser automation CLI built in Rust, designed for AI agents. It wraps Chrome via the Chrome DevTools Protocol (CDP) and exposes a fast, ergonomic command-line interface for navigation, interaction, accessibility snapshots, screenshots, network interception, and more — with no Node.js or Playwright runtime required.

Installation

Recommended (npm global)

bash
npm install -g agent-browseragent-browser install  # Download Chrome for Testing (first time only)

macOS (Homebrew)

bash
brew install agent-browseragent-browser install

Rust / Cargo

bash
cargo install agent-browseragent-browser install

Local project dependency

bash
npm install agent-browser# Add to package.json scripts or invoke via npx

Linux (with system dependencies)

bash
agent-browser install --with-deps

Quick Start

bash
agent-browser open https://example.comagent-browser snapshot                        # Accessibility tree with @refs (best for AI)agent-browser click @e2                       # Click by ref from snapshotagent-browser fill @e3 "[email protected]"   # Fill by refagent-browser get text @e1                    # Get text contentagent-browser screenshot page.pngagent-browser close

Core Commands

Navigation

bash
agent-browser open <url>           # Navigate (aliases: goto, navigate)agent-browser get url              # Get current URLagent-browser get title            # Get page titleagent-browser close                # Close browser (aliases: quit, exit)

Accessibility Snapshot (recommended for AI agents)

bash
agent-browser snapshot             # Returns accessibility tree with @ref IDsagent-browser snapshot -i          # Interactive / compact mode

Snapshot output includes @eN refs you can use directly:

@e1 [button] "Submit"@e2 [textbox] "Email" value=""@e3 [link] "Sign in"

Then act on them:

bash
agent-browser fill @e2 "[email protected]"agent-browser click @e1

Interaction

bash
agent-browser click <sel>                     # Click elementagent-browser dblclick <sel>                  # Double-clickagent-browser fill <sel> <text>               # Clear and fill inputagent-browser type <sel> <text>               # Type into elementagent-browser press <key>                     # Press key (Enter, Tab, Control+a)agent-browser keyboard type <text>            # Type at current focus (real keystrokes)agent-browser keyboard inserttext <text>      # Insert text without key eventsagent-browser hover <sel>                     # Hover elementagent-browser select <sel> <value>            # Select dropdown optionagent-browser check <sel>                     # Check checkboxagent-browser uncheck <sel>                   # Uncheck checkboxagent-browser scroll down 500                 # Scroll (up/down/left/right, optional px)agent-browser scroll down --selector "#feed"  # Scroll within elementagent-browser scrollintoview <sel>            # Scroll element into viewagent-browser drag <src> <target>             # Drag and dropagent-browser upload <sel> /path/file.pdf     # Upload file

Screenshots & PDF

bash
agent-browser screenshot                          # Save to temp dir, print pathagent-browser screenshot page.png                 # Save to pathagent-browser screenshot --full page.png          # Full-page screenshotagent-browser screenshot --annotate               # Numbered element labels overlayagent-browser screenshot --screenshot-dir ./shots # Custom output directoryagent-browser screenshot --screenshot-format jpeg --screenshot-quality 80agent-browser pdf output.pdf                      # Save page as PDF

Getting Element Info

bash
agent-browser get text <sel>           # Text contentagent-browser get html <sel>           # innerHTMLagent-browser get value <sel>          # Input valueagent-browser get attr <sel> <attr>    # Attribute valueagent-browser get count <sel>          # Count matching elementsagent-browser get box <sel>            # Bounding boxagent-browser get styles <sel>         # Computed stylesagent-browser get cdp-url              # CDP WebSocket URL

State Checks

bash
agent-browser is visible <sel>agent-browser is enabled <sel>agent-browser is checked <sel>

Semantic Locators (find)

bash
agent-browser find role button click --name "Submit"agent-browser find text "Sign In" clickagent-browser find label "Email" fill "[email protected]"agent-browser find placeholder "Search..." fill "rust"agent-browser find testid "login-btn" clickagent-browser find first ".item" clickagent-browser find nth 2 "a" textagent-browser find role textbox fill "hello" --name "Username"

Actions: click, fill, type, hover, focus, check, uncheck, text

Waiting

bash
agent-browser wait "#modal"                          # Wait for element visibleagent-browser wait 2000                              # Wait N millisecondsagent-browser wait --text "Welcome back"             # Wait for textagent-browser wait --url "**/dashboard"              # Wait for URL patternagent-browser wait --load networkidle                # Wait for load stateagent-browser wait --fn "window.appReady === true"   # Wait for JS conditionagent-browser wait "#spinner" --state hidden         # Wait for element to disappear

Load states: load, domcontentloaded, networkidle

JavaScript Eval

bash
agent-browser eval "document.title"agent-browser eval "JSON.stringify(window.__STATE__)"agent-browser eval -b "BASE64_ENCODED_JS"echo "return document.body.innerHTML" | agent-browser eval --stdin

Batch Execution (efficient multi-step)

bash
echo '[  ["open", "https://example.com"],  ["snapshot", "-i"],  ["fill", "@e2", "[email protected]"],  ["click", "@e1"],  ["screenshot", "result.png"]]' | agent-browser batch --json
# Stop on first failureagent-browser batch --bail < commands.json

Tabs & Frames

bash
agent-browser tab                    # List tabsagent-browser tab new https://...    # New tab with URLagent-browser tab 2                  # Switch to tab 2agent-browser tab close              # Close current tabagent-browser frame "#my-iframe"     # Switch into iframeagent-browser frame main             # Return to main frame

Cookies & Storage

bash
agent-browser cookiesagent-browser cookies set session_id "abc123"agent-browser cookies clear
agent-browser storage localagent-browser storage local set theme darkagent-browser storage local clearagent-browser storage session set cart '{"items":[]}'

Network

bash
agent-browser network route "**/api/users" --body '{"users":[]}'  # Mock responseagent-browser network route "**/ads/**" --abort                    # Block requestsagent-browser network unroute                                       # Remove all routesagent-browser network requests --filter api                        # View requestsagent-browser network har startagent-browser network har stop recording.har

Browser Settings

bash
agent-browser set viewport 1280 800agent-browser set viewport 375 812 2        # With device pixel ratio (retina)agent-browser set device "iPhone 14"agent-browser set geo 37.7749 -122.4194agent-browser set offline onagent-browser set headers '{"X-Custom":"value"}'agent-browser set credentials admin secretagent-browser set media dark

Auth State

bash
agent-browser state save ./auth.json    # Save cookies + localStorageagent-browser state load ./auth.json    # Restore auth stateagent-browser state list                # List saved statesagent-browser state show auth.json      # Summary of saved state

Dialogs

bash
agent-browser dialog accept             # Accept alert/confirm/promptagent-browser dialog accept "My input"  # Accept prompt with textagent-browser dialog dismiss

Clipboard

bash
agent-browser clipboard readagent-browser clipboard write "Hello, World!"agent-browser clipboard copy           # Ctrl+C current selectionagent-browser clipboard paste          # Ctrl+V

Diff & Visual Testing

bash
agent-browser diff snapshot                                  # vs last snapshotagent-browser diff snapshot --baseline before.txt            # vs saved fileagent-browser diff snapshot --selector "#main" --compactagent-browser diff screenshot --baseline before.pngagent-browser diff screenshot --baseline b.png -o diff.pngagent-browser diff url https://v1.example.com https://v2.example.comagent-browser diff url https://v1.example.com https://v2.example.com --screenshotagent-browser diff url https://v1.example.com https://v2.example.com --selector "#content"

Debug & Profiling

bash
agent-browser trace start trace.zipagent-browser trace stopagent-browser profiler startagent-browser profiler stop profile.jsonagent-browser console                  # View console messagesagent-browser errors                   # View uncaught JS exceptionsagent-browser highlight "#button"      # Visually highlight elementagent-browser inspect                  # Open Chrome DevToolsagent-browser connect 9222             # Connect to existing browser via CDP port

Common Patterns

Login flow and save session

bash
#!/bin/bashagent-browser open https://app.example.com/loginagent-browser fill "#email" "$LOGIN_EMAIL"agent-browser fill "#password" "$LOGIN_PASSWORD"agent-browser click "[type=submit]"agent-browser wait --url "**/dashboard"agent-browser state save ./session.json

AI agent loop with snapshot-driven interaction

bash
#!/bin/bashagent-browser open https://app.example.comagent-browser state load ./session.json
# Get snapshot, parse @refs, actSNAPSHOT=$(agent-browser snapshot)echo "$SNAPSHOT"
# Agent determines @e5 is the search boxagent-browser fill @e5 "quarterly report"agent-browser press Enteragent-browser wait --load networkidleagent-browser snapshotagent-browser screenshot results.png

Batch commands from a script (JSON)

bash
cat > commands.json << 'EOF'[  ["open", "https://news.ycombinator.com"],  ["wait", "--load", "networkidle"],  ["get", "title"],  ["snapshot"],  ["screenshot", "hn.png"]]EOF
agent-browser batch --json < commands.json

Scrape with mocked network

bash
agent-browser open https://api-heavy-app.example.comagent-browser network route "**/api/slow-endpoint" --body '{"data":"mocked"}'agent-browser snapshotagent-browser network unroute

Full-page screenshot with annotations

bash
agent-browser open https://example.comagent-browser wait --load networkidleagent-browser screenshot --full --annotate annotated.png

Connect to already-running Chrome

bash
# Start Chrome with remote debugginggoogle-chrome --remote-debugging-port=9222 &
agent-browser connect 9222agent-browser open https://example.comagent-browser snapshot

Emulate mobile device

bash
agent-browser set device "iPhone 14"agent-browser open https://example.comagent-browser screenshot mobile.png

HAR recording for network analysis

bash
agent-browser open https://example.comagent-browser network har startagent-browser click "#load-data"agent-browser wait --load networkidleagent-browser network har stop session.har

Selector Reference

FormatExampleNotes
@ref@e1, @e12From snapshot output — preferred for AI
CSS#id, .class, [attr=val]Standard CSS selectors
Text"Sign In"Exact text match
XPath//button[@type='submit']Full XPath

Troubleshooting

Chrome not found

bash
agent-browser install              # Downloads Chrome for Testingagent-browser install --with-deps  # Linux: also installs system libs

Element not found / timing issues

bash
agent-browser wait "#my-element"              # Wait for visibility firstagent-browser wait --load networkidle         # Wait for page to settleagent-browser wait --fn "!!document.querySelector('#app')"

Selector issues — use snapshot refs instead

bash
# Instead of fragile CSS:agent-browser click ".btn.btn-primary.submit-form"
# Use snapshot refs:agent-browser snapshot  # Find @e7 = [button] "Submit"agent-browser click @e7

Debug what's on the page

bash
agent-browser screenshot debug.png        # Visual checkagent-browser snapshot                    # Accessibility treeagent-browser console                     # JS console outputagent-browser errors                      # Uncaught exceptionsagent-browser eval "document.readyState"

Auth issues between sessions

bash
agent-browser state save ./auth.json   # After successful loginagent-browser state load ./auth.json   # At start of next session

Handling alerts/dialogs

bash
# Set up handler BEFORE the action that triggers dialogagent-browser dialog acceptagent-browser click "#delete-button"

Performance — use batch for multi-step workflows

bash
# Slow: one process per commandagent-browser open https://example.comagent-browser fill "#q" "search"agent-browser click "#submit"
# Fast: single process, multiple commandsecho '[["open","https://example.com"],["fill","#q","search"],["click","#submit"]]' \  | agent-browser batch --json

來源與署名

來源：reason-machines/trending-skills位於skills/agent-browser-automation提交2384a00

授權條款： 無授權條款

內容歸原作者所有。SourceWeft 從公開儲存庫中收錄這些內容。

檢舉或申請下架

更多來自 reason-machines/trending-skills 的技能

Yourvpndead Vpn Detection

reason-machines

介紹一款 Android 應用程式，用來偵測 VPN 用戶端並探測本機 SOCKS5 與 Clash API 連接埠是否外洩。

Security833 個月前更新

X Mentor Skill Nuwa

reason-machines

提供 X（Twitter）內容策略指引：推文與長推撰寫、鉤子評論、成長策略與帳號診斷。

Marketing & Sales833 個月前更新

Wildworld Dataset

reason-machines

介紹 WildWorld 動作條件 ARPG 世界建模資料集，並提供預期的載入、篩選與基準評測程式碼。

Data & Analytics833 個月前更新

Understand Anything Knowledge Graph

reason-machines

為程式碼庫建立可互動的知識圖譜，並支援視覺化探索、搜尋與提問。

Software Development833 個月前更新

Webphysics Avbd Engine

reason-machines

指導將實驗性 WebGPU AVBD 剛體與軟體物理引擎整合到瀏覽器專案中。

Software Development833 個月前更新

Ui Ux Pro Max Skill

reason-machines

依產品描述產生 UI/UX 設計系統，包含頁面結構、風格、配色、字體搭配與檢查清單。

Design & Creative833 個月前更新

更多Software Development技能

Playground

anthropics

精選

建立自成一體的互動式 HTML 遊樂場，包含控制項、即時預覽與可複製的提示詞輸出。

Software Development37K今天更新

M5 Onboard

anthropics

精選

透過 USB 偵測 M5Stack ESP32 開發板，燒錄 UIFlow 2.0 韌體並安裝 MicroPython 應用程式套件。

Software Development37K今天更新

Cardputer Buddy

anthropics

精選

指導迭代 Cardputer-Adv 的 MicroPython 應用程式套件：新增應用程式、透過 USB 序列埠推送檔案、查看日誌並執行 REPL 指令。

Software Development37K今天更新

Web Design Guidelines

vercel-labs

精選

依據從遠端來源取得的 Web Interface Guidelines 審查 UI 程式碼，並以 file:line 形式回報問題。

Software Development32K5 週前更新

Vercel React Native Skills

vercel-labs

精選

React Native 與 Expo 最佳實務規則，涵蓋清單效能、動畫、導覽、UI 模式、狀態與 monorepo 設定。

Software Development32K5 週前更新

Vercel React Best Practices

vercel-labs

精選

Vercel 的 React 與 Next.js 效能指南，以 70 條規則的形式用於撰寫、審查與重構程式碼。

Software Development32K5 週前更新