Loading repository data…
redf0x1 / repository
Anti-detection browser server for AI agents — REST API wrapping Camoufox engine with OpenClaw plugin support
Anti-detection browser server for AI agents — TypeScript REST API wrapping the Camoufox stealth browser engine
The Problem: Standard browser automation (Puppeteer, Playwright, Selenium) is easily detected by modern anti-bot systems. JavaScript-level patches are fragile and get bypassed quickly.
The Solution: CamoFox Browser Server wraps Camoufox, a Firefox fork with C++ engine-level fingerprint spoofing. No JavaScript injection — anti-detection happens at the browser engine level.
| Feature | Puppeteer/Playwright | CamoFox Browser Server |
|---|---|---|
| Anti-detection | JavaScript patches (fragile) | C++ engine-level (robust) |
| Fingerprint spoofing | Limited | Full (engine-level) |
| Token efficiency | Raw HTML / screenshots | Accessibility snapshots (smaller + structured) |
| Integration | Direct SDK | REST API for any language / AI agent |
| AI agent support | Varies | MCP + OpenClaw compatible |
userId (defaults: max 50 sessions, max 10 tabs/session)explicit-wins or proxy-locked)eN element references for precise interactionCAMOFOX_API_KEY is set)/start, /tabs/open, /act, etc.)CamoFox Browser Server is in Preview (Phase 1). Preview releases are functional for browser automation and agent integration, but carry specific compatibility commitments and explicit non-goals.
During Preview, CamoFox follows an additive-only deprecation model:
listItemId accepted alongside sessionKey, OpenClaw /act routing to core endpoints) continue to work alongside their replacementsBrowser profiles, download registries, and CLI session files use versioned sidecar formats. When upgrading CamoFox:
Supported sidecars include limited forward-migration paths (e.g., fingerprint v0 → v1); when no migration path exists for a given version, the server refuses to load the file and logs an actionable recovery message. There is no silent repair or downgrade path — this fail-closed default prevents data corruption at the cost of manual intervention on unsupported version jumps.
git clone https://github.com/redf0x1/camofox-browser.git
cd camofox-browser
npm install
npm run build
npm start
npm install -g camofox-browser
# Start the server
camofox-browser
# Or use the CLI for browser automation
camofox open https://example.com
camofox snapshot
camofox click e5
See CLI for the complete command reference.
Docker image:
ghcr.io/redf0x1/camofox-browser
docker build -t camofox-browser .
docker run -d \
--name camofox-browser \
-p 9377:9377 \
-p 6080:6080 \
-e CAMOFOX_HOST=0.0.0.0 \
-e CAMOFOX_AUTH_MODE=auto \
-e CAMOFOX_API_KEY=change-me \
-v ~/.camofox:/home/node/.camofox \
camofox-browser
To persist browser profiles (cookies, localStorage, IndexedDB, etc.) across container restarts, keep the volume mount shown above.
services:
camofox-browser:
build: .
ports:
- "9377:9377"
environment:
CAMOFOX_HOST: "0.0.0.0"
CAMOFOX_PORT: "9377"
# auto requires CAMOFOX_API_KEY when CAMOFOX_HOST is non-loopback
CAMOFOX_AUTH_MODE: "auto"
CAMOFOX_API_KEY: "change-me"
# CAMOFOX_ADMIN_KEY: "change-me"
# Optional: proxy routing (also enables Camoufox geoip mode)
# PROXY_HOST: ""
# PROXY_PORT: ""
# PROXY_USERNAME: ""
# PROXY_PASSWORD: ""
curl http://localhost:9377/health
# {"ok":true,"engine":"camoufox","browserConnected":true}
CamoFox Browser includes a powerful CLI for browser automation directly from the terminal. The CLI auto-starts the server when needed.
# Global install (recommended)
npm install -g camofox-browser
# Or use npx (no install needed)
npx camofox-browser open https://example.com
camofox open https://example.com # Open a page in anti-detection browser
camofox snapshot # Get accessibility tree with element refs
camofox click e5 # Click element [e5]
camofox type e3 "hello world" # Type into element [e3]
camofox screenshot --output page.png # Save screenshot
camofox close # Close the tab
# Browser lifecycle
camofox open <url> # Open URL in new tab
camofox close [tabId] # Close tab
camofox navigate <url> # Navigate current tab to URL
# Inspection
camofox snapshot # Get accessibility tree with [eN] refs
camofox screenshot [--output file] # Take screenshot (saves to file)
camofox annotate # Screenshot + element ref overlay
camofox get-url # Get current page URL
camofox get-text # Get page text content
camofox get-links # Get all links on page
camofox get-tabs # List open tabs
camofox extract-structured @schema.json # Extract deterministic JSON from a schema
# Interaction
camofox click <ref> # Click element by ref
camofox type <ref> <text> # Type text into element
camofox fill '[e1]="user" [e2]="pw"' # Fill multiple fields at once
camofox scroll <direction> # Scroll up/down/left/right
camofox select <ref> <value> # Select dropdown option
camofox hover <ref> # Hover over element
camofox press <key> # Press keyboard key
camofox drag <from> <to> # Drag element to target
# Navigation
camofox go-back # Browser back
camofox go-forward # Browser forward
camofox search "query" --engine google # Search (14 engines supported)
camofox eval "document.title" # Execute JavaScript
camofox wait <selector> [--timeout ms] # Wait for element
Text input: CamoFox has no character limit for typed or filled text. Short text stays humanized for anti-detection, while long text automatically switches to bulk DOM insertion so large inputs do not truncate.
camofox session save <name> # Save current browser state
camofox session load <name> # Restore browser state
camofox session list # List saved sessions
camofox session delete <name> # Delete saved session
camofox cookie export <file> # Export cookies to JSON file
camofox cookie import <file> # Import cookies from JSON file
Securely store credentials locally with AES-256-GCM encryption. Credentials ar
Selected from shared topics, language and repository description—not editorial ratings.
redf0x1 /
Anti-detection browser MCP server for AI agents — navigate, interact, and automate the web without getting blocked