CamoFox MCP is an anti-detection browser automation server exposing browser control tools to MCP-compatible AI agents.
Health & Status
Check server health, browser connection, and active tab counts (
server_status)Retrieve session statistics: request counts, active tabs, uptime, and performance metrics (
get_stats)
Tab Management
Create browser tabs with anti-detection fingerprinting and optional user session isolation (
create_tab)Close tabs to free resources (
close_tab)List all open tabs with URLs and titles (
list_tabs)
Navigation
Navigate to URLs with full page load waiting (
navigate)Go back/forward in browser history (
go_back,go_forward)Refresh the current page (
refresh)
Interaction
Click elements via accessibility tree refs or CSS selectors (
click)Type text into input fields (
type_text)Scroll pages by a specified pixel amount (
scroll)Press keyboard keys (Enter, Tab, Escape, Arrow keys, etc.) (
press_key)
Observation & Content Extraction
Capture a token-efficient accessibility tree snapshot with element refs for interaction (
snapshot)Take visual screenshots as base64 PNG for layout verification (
screenshot)Extract all hyperlinks with URLs and link text (
get_links)
Web Search
Search across 14 engines: Google, YouTube, Amazon, Bing, DuckDuckGo, Reddit, GitHub, StackOverflow, Wikipedia, Twitter, LinkedIn, Facebook, Instagram, and TikTok (
web_search)
Session Management
Import cookies as JSON to restore authenticated sessions without re-authentication (
import_cookies)
Primary use cases: web scraping with anti-detection, automated form filling, multi-account management with session isolation, search result gathering, and authenticated workflow automation.
Provides a built-in search macro to search Amazon using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search DuckDuckGo using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search Facebook using anti-detection technology to avoid being blocked.
Utilizes a Firefox-based anti-detection browser engine to navigate, interact, and automate the web with unique, human-like fingerprints.
Provides a built-in search macro to search GitHub using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search Google using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search Instagram using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search Reddit using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search TikTok using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search Wikipedia using anti-detection technology to avoid being blocked.
Provides a built-in search macro to search YouTube using anti-detection technology to avoid being blocked.
CamoFox MCP
AI-powered anti-detection browser automation for MCP-compatible AI agents.
CamoFox MCP connects MCP clients such as Claude Desktop, VS Code, Cursor, and OpenClaw to the CamoFox browser server. It gives agents a practical browser toolset for navigation, interaction, search, extraction, downloads, and session reuse while relying on Camoufox-based anti-detection behavior underneath.
Key Features
46 browser automation tools across navigation, interaction, observation, search, downloads, sessions, and batch workflows.
Anti-detection browser automation built on top of the CamoFox browser server and Camoufox.
Multi-tab workflows with tracked state, history, and cleanup.
Session persistence with cookie import, saved profiles, and optional auto-save.
Token-efficient accessibility snapshots with CSS-selector fallbacks for difficult SPA flows.
OpenClaw-compatible HTTP transport, plus standard stdio support for desktop MCP clients.
Quick Install
You need both components running:
camofox-browserhandles the anti-detection browser.camofox-mcpexposes that browser to your MCP client.
Option A: npx + stdio
Start the browser server:
npx camofox-browser@latestAdd CamoFox MCP to your MCP client:
{
"servers": {
"camofox": {
"type": "stdio",
"command": "npx",
"args": ["-y", "camofox-mcp@latest"],
"env": {
"CAMOFOX_URL": "http://localhost:9377"
}
}
}
}Option B: Docker
Start the browser server:
docker run -d -p 9377:9377 --name camofox-browser ghcr.io/redf0x1/camofox-browser:latestRun CamoFox MCP in HTTP mode for remote MCP clients such as OpenClaw:
docker run -p 3000:3000 --rm \
-e CAMOFOX_TRANSPORT=http \
-e CAMOFOX_URL=http://host.docker.internal:9377 \
ghcr.io/redf0x1/camofox-mcp:latest node dist/http.jsFull client configuration examples live in docs/getting-started.md.
Quick Verify
Verify the browser server is reachable:
curl -fsS http://localhost:9377/healthExpected response includes "ok":true and "browserConnected":true.
Tool Categories
Category | Tool count | Docs |
Health | 1 | |
Tabs | 3 | |
Navigation | 4 | |
Interaction | 8 | |
Observation | 8 | |
Search | 1 | |
Session | 4 | |
Profiles | 4 | |
Downloads | 3 | |
Extraction | 3 | |
Batch workflows | 6 | |
Presets | 1 |
Top Limitations
CamoFox MCP is not a standalone browser. You must run a compatible
camofox-browserserver separately.Accessibility-tree refs are the primary interaction model, but SPA and custom-component sites can require CSS selectors or rendered HTML tools.
If the browser server enforces authentication, API-key-gated operations need the same
CAMOFOX_API_KEYon both sides.HTTP transport is mainly for remote MCP clients. Desktop MCP clients usually work best with stdio configuration.
Security
Treat this as a browser control surface. In shared or networked environments, isolate the browser server, avoid exposing MCP endpoints broadly, and use CAMOFOX_API_KEY when authentication is enabled. Session profiles can contain sensitive cookies and should be stored accordingly.
Documentation
Start at docs/README.md for the documentation hub, then use docs/getting-started.md for setup, verification, and first workflow examples.
Contributing + License
See CONTRIBUTING.md for contribution guidelines and LICENSE for the MIT license.