Skip to content

Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.

8.3 KB
Raw Blame

Integrations

Table of Contents


Browser Use toolsets for Claude

Browser Use toolsets for Claude is maintained by Browser Use and is compatible with Claude. Anthropic's SDK owns the model loop and tool runner. Browser Use implements the 31 browser actions and connects them to local Chromium, Browser Use Cloud, or an existing browser over CDP.

Requirements:

  • Python 3.11 or newer
  • Linux or macOS with /bin/bash, or WSL on Windows
  • An Anthropic SDK release with anthropic.tools.browser and client.beta.messages.tool_runner
  • ANTHROPIC_API_KEY and the model ID Anthropic documents for the browser toolset
  • BROWSER_USE_API_KEY only when using Browser Use Cloud. Create one at cloud.browser-use.com/new-api-key.

Install the packages and local Chromium:

uv init --python 3.12
uv add browser-use anthropic
uvx browser-use install

Create run_browser.py:

import asyncio
from pathlib import Path

from anthropic import AsyncAnthropic

from browser_use.integrations.toolsets_for_claude import Bash, BrowserUse


async def main() -> None:
    task = 'Open example.com and save its page title to title.txt.'
    driver = BrowserUse()
    # driver = BrowserUse(use_cloud=True)  # Requires BROWSER_USE_API_KEY
    bash = Bash(output_dir=Path('outputs'))

    async with driver, AsyncAnthropic() as client:
        runner = client.beta.messages.tool_runner(
            model='claude-opus-5-5',
            max_tokens=32_768,
            max_iterations=1_000,
            tools=[driver, bash],
            messages=[{'role': 'user', 'content': task}],
        )
        final = await runner.until_done()

    print('\n'.join(block.text for block in final.content if block.type == 'text'))


asyncio.run(main())

Set the Anthropic variables and run it:

export ANTHROPIC_API_KEY=your-key
# Optional SDK request and tool-runner logs
export ANTHROPIC_LOG=info

uv run run_browser.py

The async with driver block starts and closes a driver-owned local or Cloud browser. When passing an already-started BrowserSession, the application starts and closes that session instead.

BrowserUse implements all 31 actions. Anthropic leaves javascript_exec, file_upload, read_console, and read_network disabled by default. Enable only the capabilities the application needs:

async def confirm(context):
    if context.member not in {'javascript_exec', 'file_upload'}:
        return True
    return await app.approve(context.member, context.tab_url)


driver = BrowserUse(
    confirm=confirm,
    configs={
        'javascript_exec': {'enabled': True},
        'file_upload': {'enabled': True},
        'read_console': {'enabled': True},
        'read_network': {'enabled': True},
    },
)

Bash is a separate bounded tool for local computation and deliverables:

bash = Bash(output_dir='outputs', timeout_seconds=120, max_output_bytes=50_000)

Bash requires Linux or macOS with /bin/bash, or WSL on Windows. It strips ambient credentials from child commands and limits execution time and returned output. Its working directory is the default location for commands, which can access other files available to the process; it is not an operating-system sandbox. Run the SDK process in your normal container or sandbox for untrusted tasks.

For a remote browser, upload paths must already exist on the browser host. Local files created by Bash are not copied to that host automatically. Use the SDK's file policy and a document_resolver that maps approved document IDs to browser-host paths.

See the complete quickstart, runtime examples, action list, and file guidance.


MCP Server (Cloud)

HTTP-based MCP server at https://api.browser-use.com/mcp

Setup

Claude Code:

claude mcp add --transport http browser-use https://api.browser-use.com/mcp

Claude Desktop (macOS ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "browser-use": {
      "type": "http",
      "url": "https://api.browser-use.com/mcp",
      "headers": { "x-browser-use-api-key": "your-api-key" }
    }
  }
}

Cursor (~/.cursor/mcp.json):

{
  "mcpServers": {
    "browser-use": {
      "type": "http",
      "url": "https://api.browser-use.com/mcp",
      "headers": { "x-browser-use-api-key": "your-api-key" }
    }
  }
}

Windsurf (~/.codeium/windsurf/mcp_config.json):

{
  "mcpServers": {
    "browser-use": {
      "type": "http",
      "url": "https://api.browser-use.com/mcp",
      "headers": { "x-browser-use-api-key": "your-api-key" }
    }
  }
}

Cloud MCP Tools

Tool Cost Description
browser_task $0.01 + per-step Run browser automation task
execute_skill $0.02 Execute a skill
list_skills Free List available skills
get_cookies Free Get cookies
list_browser_profiles Free List cloud profiles
monitor_task Free Check task progress

browser_task params: task (required), max_steps (1-10, default 8), profile_id (UUID)


MCP Server (Local)

Free, self-hosted stdio-based server:

uvx --from 'browser-use[cli]' browser-use --mcp

Claude Desktop Config

macOS (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "browser-use": {
      "command": "/Users/your-username/.local/bin/uvx",
      "args": ["--from", "browser-use[cli]", "browser-use", "--mcp"],
      "env": {
        "OPENAI_API_KEY": "your-key"
      }
    }
  }
}

Note: Use full path to uvx on macOS/Linux (run which uvx to find it).

Local MCP Tools

Agent: retry_with_browser_use_agent — full automation task

Direct Control:

  • browser_navigate — Go to URL
  • browser_click — Click element by index
  • browser_type — Type text
  • browser_get_state — Page state + interactive elements
  • browser_scroll — Scroll page
  • browser_go_back — Back in history

Tabs: browser_list_tabs, browser_switch_tab, browser_close_tab

Extraction: browser_extract_content — Structured extraction

Sessions: browser_list_sessions, browser_close_session, browser_close_all

Environment Variables

  • OPENAI_API_KEY or ANTHROPIC_API_KEY — LLM key (required)
  • BROWSER_USE_HEADLESS — false to show browser
  • BROWSER_USE_DISABLE_SECURITY — true to disable security
  • BROWSER_USE_LOGGING_LEVEL — DEBUG for verbose logs

Programmatic Usage

from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client

async def use_browser_mcp():
    server_params = StdioServerParameters(
        command="uvx",
        args=["--from", "browser-use[cli]", "browser-use", "--mcp"]
    )
    async with stdio_client(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()
            result = await session.call_tool("browser_navigate", arguments={"url": "https://example.com"})

Skills

Load cloud skills into agents as reusable API endpoints:

agent = Agent(
    task='Analyze TikTok and Instagram profiles',
    skills=[
        'a582eb44-e4e2-4c55-acc2-2f5a875e35e9',  # TikTok Scraper
        'f8d91c2a-3b4e-4f7d-9a1e-6c8e2d3f4a5b',  # Instagram Scraper
    ],
    llm=ChatBrowserUse()
)
await agent.run()
  • Use skills=['*'] for all skills (each adds ~200 tokens to prompt)
  • Requires BROWSER_USE_API_KEY
  • Browse/create at cloud.browser-use.com/skills
  • Cookies auto-injected from browser; if missing, LLM navigates to obtain them

Documentation MCP

Read-only docs access (no browser automation):

Claude Code:

claude mcp add --transport http browser-use-docs https://docs.browser-use.com/mcp

Cursor (~/.cursor/mcp.json):

{
  "mcpServers": {
    "browser-use-docs": { "url": "https://docs.browser-use.com/mcp" }
  }
}

No API key needed. Provides API reference, config options, best practices, examples.