Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@camoufox-mcp navigate to example.com and take a screenshot"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server for browser automation using Camoufox anti-detect browser.
Anti-detect browser with BrowserForge fingerprinting
Automatic geolocation/timezone spoofing from proxy IP
Human-like cursor movement
DOM snapshot-based element targeting
Related MCP server: camofox-mcp
Requires uv (handles Python 3.14+ automatically).
Add to .cursor/mcp.json:
{
"mcpServers": {
"camoufox": {
"command": "uvx",
"args": ["camoufox-mcp"]
}
}
}{
"mcpServers": {
"camoufox": {
"command": "uvx",
"args": [
"camoufox-mcp",
"--proxy", "http://user:pass@proxy.example.com:8080"
]
}
}
}Camoufox automatically spoofs geolocation and timezone based on the proxy IP.
{
"mcpServers": {
"camoufox": {
"command": "uvx",
"args": ["camoufox-mcp", "--headed", "--humanize", "1.5"]
}
}
}Option | Description |
| Show browser window |
| Proxy server (http/socks5) |
| Disable geolocation spoofing |
| Browser locale (e.g., en-US) |
| Cursor movement intensity (0-2) |
| Block image loading |
Tool | Description |
| Navigate to URL |
| Go back in history |
| Get page structure with element refs |
| Click element by ref |
| Type text into element |
| Fill text (clears first) |
| Hover over element |
| Select dropdown option |
| Press keyboard key |
| Take screenshot |
| Manage tabs |
| Wait for text/time |
| Run JavaScript |
| Close browser |
MIT
Click an element by its ref from the snapshot
Inputs: ref (required), element.
Close the browser
Inputs: no named parameters.
Evaluate JavaScript in the page context
Inputs: expression (required).
Fill text into an element (clears first)
Inputs: ref (required), text (required).
Hover over an element
Inputs: ref (required).
Press a keyboard key
Inputs: key (required).
Take a screenshot of the current page
Inputs: full_page.
Select option(s) in a dropdown
Inputs: ref (required), values (required).
Get accessibility snapshot of the current page. Returns element refs (e.g., e0, e1) for interactions.
Inputs: no named parameters.
Manage browser tabs: list, new, close, or select
Inputs: action (required), index.
Type text into an element (appends to existing)
Inputs: ref (required), text (required), submit.
Wait for text or a specified time
Inputs: text, time, text_gone.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 14 tools
Each tool has a clearly distinct action: navigate, snapshot, click, type, fill, hover, select, key press, wait, evaluate, screenshot, tabs, back, close. The only potential overlap is type vs. fill, but descriptions clarify that fill clears first, making them distinct.
All tools use the 'browser_' prefix followed by a verb_noun pattern (e.g., browser_navigate, browser_click, browser_screenshot). The naming is predictable and uniform, with only minor deviations like browser_wait_for (verb_preposition) and browser_tabs (verb_noun but plural) that are still clear.
14 tools is an ideal size for a browser automation server. It covers all essential interactions without being bloated. Each tool serves a specific purpose and there are no redundant or unnecessary tools.
The tool set covers the core browser automation lifecycle: navigation, interaction, tab management, screenshots, and JavaScript evaluation. Minor gaps exist, such as the absence of navigate_forward (only back) and a dedicated wait-for-element function (only wait for text or time). These are small but notable omissions.
Web scraping MCP server — scrape, extract structured data, screenshot any site with anti-bot bypass.
Official Browserless MCP server: browser automation, scraping, screenshots and PDFs for AI agents.
Hosted Web Crawler MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
Zenrows MCP server: Fetch, Extract, Batch, and Browser Sessions for AI agents
