메뉴
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, or extract information from web pages.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Manages issues via the br CLI (beads), a local-first tracker stored in .beads/ (SQLite + JSONL). Use when capturing issues, finding ready work, claiming issues, managing dependencies, or syncing changes. Triggers: "br", "beads", "issue tracking".
Review, rewrite, and debug LLM prompts. Use when asked to: (1) improve an existing prompt, (2) design a new system or developer prompt, (3) add few-shot examples, (4) define output schemas and success criteria, (5) diagnose prompt failures from transcripts. Not for: model policy workarounds, producing domain answers without context, or guaranteeing model-specific behavior.
Creates or updates mala.yaml project configuration files for the local mala orchestrator. Use when asked to initialize, generate, write, edit, validate, fix, or explain a mala.yaml config, including presets, commands, coverage, evidence checks, validation triggers, or review settings.
Simplifies and refines code for clarity, consistency, and maintainability while preserving all functionality. Focuses on recently modified code unless instructed otherwise. Triggers: simplify, refine, clean up code.
Executes SQL queries on Dune Analytics and writes results as CSV to /tmp. Use when the user needs blockchain data from Dune or wants to run a saved Dune query by ID. Triggers: "dune query", "dune sql", "query dune", "blockchain data"
Explore Python import graphs and enforce layered architecture using grimp.
| name | browser-use |
| description | Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, or extract information from web pages. |
| allowed-tools | Bash(uvx browser-use:*) |
The browser-use command provides fast, persistent browser automation. It maintains browser sessions across commands, enabling complex multi-step workflows.
uvx browser-use open https://example.com # Navigate to URL
uvx browser-use state # Get page elements with indices
uvx browser-use click 5 # Click element by index
uvx browser-use type "Hello World" # Type text
uvx browser-use screenshot # Take screenshot
uvx browser-use close # Close browser
uvx browser-use open <url> - Opens URL (starts browser if needed)uvx browser-use state - Returns clickable elements with indicesuvx browser-use click 5, uvx browser-use input 3 "text")uvx browser-use state or uvx browser-use screenshot to confirm actionsuvx browser-use --browser chromium open <url> # Default: headless Chromium
uvx browser-use --browser chromium --headed open <url> # Visible Chromium window
uvx browser-use --browser real open <url> # User's Chrome with login sessions
uvx browser-use --browser remote open <url> # Cloud browser (requires API key)
uvx browser-use open <url> # Navigate to URL
uvx browser-use back # Go back in history
uvx browser-use scroll down # Scroll down
uvx browser-use scroll up # Scroll up
uvx browser-use state # Get URL, title, and clickable elements
uvx browser-use screenshot # Take screenshot (outputs base64)
uvx browser-use screenshot path.png # Save screenshot to file
uvx browser-use screenshot --full path.png # Full page screenshot
uvx browser-use state)uvx browser-use click <index> # Click element
uvx browser-use type "text" # Type text into focused element
uvx browser-use input <index> "text" # Click element, then type text
uvx browser-use keys "Enter" # Send keyboard keys
uvx browser-use keys "Control+a" # Send key combination
uvx browser-use select <index> "option" # Select dropdown option
uvx browser-use switch <tab> # Switch to tab by index
uvx browser-use close-tab # Close current tab
uvx browser-use close-tab <tab> # Close specific tab
uvx browser-use eval "document.title" # Execute JavaScript, return result
uvx browser-use extract "all product prices" # Extract data using LLM (requires API key)
uvx browser-use python "x = 42" # Set variable
uvx browser-use python "print(x)" # Access variable (outputs: 42)
uvx browser-use python "print(browser.url)" # Access browser object
uvx browser-use python --vars # Show defined variables
uvx browser-use python --reset # Clear Python namespace
uvx browser-use python --file script.py # Execute Python file
The Python session maintains state across commands. The browser object provides:
browser.url - Current page URLbrowser.title - Page titlebrowser.goto(url) - Navigatebrowser.click(index) - Click elementbrowser.type(text) - Type textbrowser.screenshot(path) - Take screenshotbrowser.scroll() - Scroll pagebrowser.html - Get page HTMLuvx browser-use run "Fill the contact form with test data" # Run AI agent
uvx browser-use run "Extract all product prices" --max-steps 50
Agent tasks use an LLM to autonomously complete complex browser tasks. Requires BROWSER_USE_API_KEY or configured LLM API key (OPENAI_API_KEY, ANTHROPIC_API_KEY, etc).
uvx browser-use sessions # List active sessions
uvx browser-use close # Close current session
uvx browser-use close --all # Close all sessions
uvx browser-use server status # Check if server is running
uvx browser-use server stop # Stop server
uvx browser-use server logs # View server logs
| Option | Description |
|---|---|
--session NAME | Use named session (default: "default") |
--browser MODE | Browser mode: chromium, real, remote |
--headed | Show browser window (chromium mode) |
--profile NAME | Chrome profile (real mode only) |
--json | Output as JSON |
--api-key KEY | Override API key |
Session behavior: All commands without --session use the same "default" session. The browser stays open and is reused across commands. Use --session NAME to run multiple browsers in parallel.
uvx browser-use open https://example.com/contact
uvx browser-use state
# Shows: [0] input "Name", [1] input "Email", [2] textarea "Message", [3] button "Submit"
uvx browser-use input 0 "John Doe"
uvx browser-use input 1 "john@example.com"
uvx browser-use input 2 "Hello, this is a test message."
uvx browser-use click 3
uvx browser-use state # Verify success
uvx browser-use --session work open https://work.example.com
uvx browser-use --session personal open https://personal.example.com
uvx browser-use --session work state # Check work session
uvx browser-use --session personal state # Check personal session
uvx browser-use close --all # Close both sessions
uvx browser-use open https://example.com/products
uvx browser-use python "
products = []
for i in range(20):
browser.scroll('down')
browser.screenshot('products.png')
"
uvx browser-use python "print(f'Captured {len(products)} products')"
uvx browser-use --browser real open https://gmail.com
# Uses your actual Chrome with existing login sessions
uvx browser-use state # Already logged in!
uvx browser-use state first to see available elements and their indices--headed for debugging to see what the browser is doing--json for parsing output programmaticallyuvx browser-use python commands within a sessionBrowser won't start?
uvx browser-use server stop # Stop any stuck server
uvx browser-use --headed open <url> # Try with visible window
Element not found?
uvx browser-use state # Check current elements
uvx browser-use scroll down # Element might be below fold
uvx browser-use state # Check again
Session issues?
uvx browser-use sessions # Check active sessions
uvx browser-use close --all # Clean slate
uvx browser-use open <url> # Fresh start
Always close the browser when done. Run this after completing browser automation:
uvx browser-use close