browser-use-mcp/README.md

Official Browser Use MCP server for real-time web data access via the Browser Use API.

# browser-use MCP Server

Welcome to the official Model Context Protocol (MCP) server for [browser-use.com](https://browser-use.com), enabling LLMs, agents, and apps to access, search, and extract web data in real-time using the browser-use.com API. This server exposes MCP tools for web search and concurrent URL search, making it easy to integrate real web data into any MCP-compatible client (like Claude Desktop, Cursor, Windsurf, and more).

## 🔥 Why browser-use Over Traditional Web Search APIs?

**Get live data, not cached results.** Unlike traditional web search APIs that return indexed data, browser-use visits websites in real-time:

- 🚀 **Always current**: Live prices, breaking news, real-time analytics
- 🎯 **No stale data**: See exactly what's on the page right now
- 🌐 **Dynamic content**: Access JavaScript-rendered sites and interactive dashboards

**When you need NOW, not "last indexed 3 days ago" - choose browser-use.**

## 🚀 Quick Start

### Claude Desktop

1. Get your API key from [cloud.browser-use.com/billing](https://cloud.browser-use.com/billing)
2. Add to Claude Desktop config (`Settings > Developer > Edit Config`):

```json
{
    "mcpServers": {
        "browser_use": {
            "command": "npx",
            "args": ["-y", "browser-use-mcp"],
            "env": {
                "BROWSER_USE_API_KEY": "YOUR-BROWSERUSE-API-KEY-HERE"
            }
        }
    }
}
```

3. Restart Claude Desktop
4. Test with: "What is the current the stock price of Alphabet?"

### Other MCP Clients

```sh
export BROWSER_USE_API_KEY=your-api-key
npx browser-use-mcp
```

## 🔧 Available Tools

### `search-web`
Search the web with live results
- **Input**: `{ "query": "search string", "max_websites": 3, "depth": 2 }`
- **Returns**: JSON search results
- `max_websites`: Optional (default: 3, max: 6)
- `depth`: Optional (default: 2, range: 2-5) - How deep to navigate within each website

### `search-urls`
Search multiple URLs concurrently
- **Input**: `{ "urls": ["url1", "url2"], "query": "search string", "depth": 2 }`
- **Returns**: Array of results for each URL
- Maximum 10 URLs per request
- `depth`: Optional (default: 2, range: 2-5) - How deep to navigate within each website

**Depth explanation:**
- `depth=2`: Checks main page + 1 click deeper
- `depth=3`: Checks main page + 2 clicks deeper  

## 🛠️ Setup

1. 🌐 Create account at [cloud.browser-use.com](https://cloud.browser-use.com)
2. 🔑 Generate API key on the billing page
3. 💻 Set `BROWSER_USE_API_KEY` environment variable

## ⚠️ Security Best Practices

- Treat all scraped content as untrusted
- Validate web data before processing
- Avoid using raw content directly in prompts (prompt injection risk)

## Support

For help, visit [browser-use.com](https://browser-use.com) or open a GitHub issue.

---

**Keywords**: mcp, browser-use, web scraping, web search, web automation, mcp-server