camoufox-mcp
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 Go to google.com and search for 'camoufox browser'"
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.
Camoufox MCP Server
Deprecated: moved to camoufox-browser
This repository is deprecated. The project has moved to:
Repository: https://github.com/rlgrpe/camoufox-browser-cli
A Python MCP (Model Context Protocol) server providing full browser automation with Camoufox anti-detection capabilities.
Features
21 browser automation tools for complete browser control
Anti-detection: Automatic fingerprint randomization to avoid bot detection
Lazy initialization: Browser only launches on first tool call
Persistent sessions: Browser persists between tool calls
Tab management: Support for multiple tabs/pages
Form automation: Fill forms, upload files, handle dialogs
Related MCP server: Camoufox Browser MCP
Installation
Quick Start (uvx - recommended)
uvx camoufox-mcpUsing pipx
pipx run camoufox-mcpUsing pip
pip install camoufox-mcp
camoufox-mcpFrom source
# Clone the repository
git clone https://github.com/rlgrpe/camoufox-mcp-python.git
cd camoufox-mcp-python
# Install with uv
uv syncInstall Camoufox browser
After installing the package, you need to download the Camoufox browser binary. The command depends on how you installed the package:
If installed via uvx (recommended)
uvx camoufox fetchIf installed via pipx
pipx run camoufox fetchIf installed via pip
# macOS
python3 -m camoufox fetch
# Linux
python -m camoufox fetch
# Windows
camoufox fetchIf installed from source (uv)
uv run python -m camoufox fetchLinux system dependencies
On Linux, you also need to install system dependencies:
# Ubuntu/Debian
sudo apt install -y libgtk-3-0 libx11-xcb1 libasound2
# Arch
sudo pacman -S gtk3 libx11 libxcb cairo libasound alsa-libTo uninstall the browser binary
# Use the same method as installation:
uvx camoufox remove # if using uvx
pipx run camoufox remove # if using pipx
camoufox remove # if using pip
uv run python -m camoufox remove # if using uv from sourceUsage
Start the server
# Using uvx (recommended)
uvx camoufox-mcp
# Using uv (from source)
uv run camoufox-mcp
# Or as a module
python -m camoufox_mcpAdd to Claude Code
claude mcp add camoufox-mcp -- uvx camoufox-mcpClaude Desktop Configuration
Add to your Claude Desktop config file:
{
"mcpServers": {
"camoufox-mcp": {
"command": "uvx",
"args": ["camoufox-mcp"]
}
}
}Or if running from source:
{
"mcpServers": {
"camoufox-mcp": {
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/camoufox-mcp-python",
"camoufox-mcp"
]
}
}
}Tools Reference
Navigation (4 tools)
Tool | Description |
| Navigate to a URL |
| Go back to the previous page |
| Capture accessibility snapshot of the current page |
| Wait for text to appear/disappear or a specified time |
Interaction (6 tools)
Tool | Description |
| Click on an element |
| Type text into an element |
| Hover over an element |
| Drag and drop between two elements |
| Press a key on the keyboard |
| Select an option in a dropdown |
Page (4 tools)
Tool | Description |
| Evaluate JavaScript on the page or element |
| Get console messages |
| Get network requests |
| Take a screenshot |
Forms (3 tools)
Tool | Description |
| Fill multiple form fields |
| Upload files |
| Handle dialogs (alert, confirm, prompt) |
Session (4 tools)
Tool | Description |
| List, create, close, or select tabs |
| Resize the browser window |
| Close the browser |
| Install Camoufox browser |
Configuration
The browser can be configured via the BrowserConfig class in camoufox_mcp/config.py. Available options:
Parameter | Type | Default | Description |
| str | None | OS to spoof (windows/macos/linux) |
| bool/str | False | Headless mode (or 'virtual') |
| bool/float | True | Human-like cursor movements |
| bool | True | Auto-detect geolocation from IP |
| str/dict | None | Proxy configuration |
| bool | True | Block WebRTC leaks |
| bool | False | Block WebGL fingerprinting |
| bool | False | Block image loading |
| str | None | Browser locale |
| dict | None | Viewport dimensions |
| bool | True | Enable caching |
| int | 30000 | Page load timeout (ms) |
| str | "domcontentloaded" | Wait strategy |
| bool | False | Enable uBlock Origin |
Proxy Configuration Example
proxy = {
'server': 'http://proxy.example.com:8080',
'username': 'user',
'password': 'pass'
}Development
# Install dev dependencies
uv sync --dev
# Run tests
uv run pytest
# Format code
uv run ruff format
# Lint
uv run ruff checkLicense
MIT License - see LICENSE for details.
Related Projects
Available Tools
22 toolsbrowser_clickA
Perform click on a web page element.
Args:
element: Human-readable element description
ref: Exact target element reference from the page snapshot
button: Button to click (left, right, middle). Defaults to left.
double_click: Whether to perform a double click
Returns:
Success message
Inputs: element (required), ref (required), button, double_click. Returns: result
browser_closeA
Close the browser.
This will terminate the browser instance and clean up all resources.
A new browser will be launched on the next navigation command.
Returns:
Success message
Inputs: no named parameters. Returns: result
browser_configureA
Configure proxy before launching browser.
Call BEFORE browser_navigate. If browser running, call browser_close first.
Args:
proxy: Supports formats:
- login:password@ip:port
- ip:port:login:password
- http://login:password@ip:port
- ip:port
Returns:
Status message
Inputs: proxy. Returns: result
browser_console_messagesB
Returns all console messages.
Args:
level: Minimum level of messages to return. Each level includes
messages of more severe levels. Defaults to "info".
Returns:
Formatted console messages
Inputs: level. Returns: result
browser_dragB
Perform drag and drop between two elements.
Args:
start_element: Human-readable source element description
start_ref: Exact source element reference from the page snapshot
end_element: Human-readable target element description
end_ref: Exact target element reference from the page snapshot
Returns:
Success message
Inputs: start_element (required), start_ref (required), end_element (required), end_ref (required). Returns: result
browser_evaluateA
Evaluate JavaScript expression on page or element.
Args:
function: JavaScript function to execute. Format:
'() => { /* code */ }' or
'(element) => { /* code */ }' when element is provided
element: Human-readable element description (optional)
ref: Exact target element reference for element-scoped evaluation (optional)
Returns:
Result of the JavaScript evaluation
Inputs: function (required), element, ref. Returns: result
browser_file_uploadA
Upload one or multiple files.
This tool should be called when a file chooser dialog appears.
If paths is omitted, the file chooser is cancelled.
Args:
paths: The absolute paths to the files to upload. Can be single
or multiple files. If omitted, file chooser is cancelled.
Returns:
Success message
Inputs: paths. Returns: result
browser_fill_formA
Fill multiple form fields.
Args:
fields: List of form fields to fill. Each field should have:
- name: Human-readable field name
- ref: Exact target field reference from the page snapshot
- type: Field type (textbox, checkbox, radio, combobox, slider)
- value: Value to fill (for checkbox use 'true'/'false',
for combobox use option text)
Returns:
Success message with filled fields
Inputs: fields (required). Returns: result
browser_handle_dialogA
Handle a dialog (alert, confirm, prompt, beforeunload).
Args:
accept: Whether to accept the dialog.
prompt_text: The text of the prompt in case of a prompt dialog.
Returns:
Success message
Inputs: accept (required), prompt_text. Returns: result
browser_hoverC
Hover over an element on the page.
Args:
element: Human-readable element description
ref: Exact target element reference from the page snapshot
Returns:
Success message
Inputs: element (required), ref (required). Returns: result
browser_installA
Install Camoufox browser.
Call this if you get an error about the browser not being installed.
This will download and install the Camoufox browser binary.
Returns:
Installation result
Inputs: no named parameters. Returns: result
browser_network_requestsA
Returns all network requests since loading the page.
Args:
include_static: Whether to include static resources like images,
fonts, scripts, etc. Defaults to false.
Returns:
Formatted network requests
Inputs: include_static. Returns: result
browser_press_keyB
Press a key on the keyboard.
Args:
key: Name of the key to press (e.g., 'ArrowLeft', 'Enter', 'a')
Returns:
Success message
Inputs: key (required). Returns: result
browser_resizeB
Resize the browser window.
Args:
width: Width of the browser window in pixels
height: Height of the browser window in pixels
Returns:
Success message
Inputs: width (required), height (required). Returns: result
browser_select_optionA
Select an option in a dropdown.
Args:
element: Human-readable element description
ref: Exact target element reference from the page snapshot
values: Array of values to select (can be single or multiple)
Returns:
Success message
Inputs: element (required), ref (required), values (required). Returns: result
browser_snapshotA
Capture accessibility snapshot of the current page.
This provides a structured view of the page content that is better
for understanding page structure than screenshots.
Returns:
Accessibility tree of the current page
Inputs: no named parameters. Returns: result
browser_tabsA
List, create, close, or select a browser tab.
Args:
action: Operation to perform:
- list: List all open tabs
- new: Create a new tab
- close: Close a tab (uses index if provided, else current tab)
- select: Select a tab by index
index: Tab index, used for close/select operations.
If omitted for close, current tab is closed.
Returns:
Result of the tab operation
Inputs: action (required), index. Returns: result
browser_take_screenshotA
Take a screenshot of the current page or element.
Args:
full_page: When true, takes a screenshot of the full scrollable page.
Cannot be used with element screenshots.
filename: File name to save the screenshot to. Defaults to
'page-{timestamp}.{png|jpeg}' if not specified.
element: Human-readable element description (optional)
ref: Exact target element reference for element screenshot (optional)
type: Image format for the screenshot (png or jpeg). Defaults to png.
Returns:
Path to the saved screenshot
Inputs: full_page, filename, element, ref, +1 more. Returns: result
browser_typeA
Type text into an editable element.
Args:
element: Human-readable element description
ref: Exact target element reference from the page snapshot
text: Text to type into the element
slowly: Whether to type one character at a time (useful for key handlers)
submit: Whether to submit entered text (press Enter after)
Returns:
Success message
Inputs: element (required), ref (required), text (required), slowly, +1 more. Returns: result
browser_wait_forB
Wait for text to appear, disappear, or a specified time to pass.
Args:
text: Text to wait for to appear
text_gone: Text to wait for to disappear
time: Time to wait in seconds
Returns:
Success message when condition is met
Inputs: text, text_gone, time. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 22 tools
Each tool targets a distinct browser automation action (navigate, click, type, screenshot, etc.) with no overlap. Descriptions clearly differentiate them.
All tools follow a consistent 'browser_verb' pattern (e.g., browser_click, browser_navigate, browser_take_screenshot). The naming is uniform and predictable.
22 tools is slightly above the typical 3-15 range, but the comprehensive coverage of browser automation operations justifies the count. Not excessive.
Covers all major browser interactions: navigation, clicking, typing, form filling, dropdowns, file upload, dialogs, tabs, screenshots, snapshots, console/network, JS evaluation, waiting, resizing, proxy config, install, close. No obvious gaps.
Maintenance
Related MCP Connectors
Undetectable cloud browser sessions for AI agents and scrapers. Navigate, extract, click, captcha.
Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data.
Stealth web automation for AI agents. Login, signup, navigate, screenshot.
Stealth web automation for AI agents. Login, signup, navigate, screenshot.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables browser automation and web scraping with multi-session management, supporting page navigation, element interaction, network request capture, and content extraction across multiple concurrent browser instances.MIT
- AlicenseNot gradedqualityDmaintenanceProvides stealthy browser automation capabilities using a custom Firefox build designed for anti-detection. It enables users to navigate pages, interact with elements, and manage persistent browser sessions through natural language.1ISC
- AlicenseAqualityCmaintenanceEnables AI agents to fully control a browser for web automation, including navigation, clicking, typing, scrolling, screenshots, and DOM inspection, with session persistence and anti-bot bypass.4036 npmMIT
- AlicenseCqualityDmaintenanceEnables AI-powered browser automation with real profiles, supporting interaction, navigation, form filling, and advanced features like visual overlays and self-healing selectors.42Apache 2.0