stealth-browser-mcp MCP — Cas d'usage, Installer & Démo en direct

Pourquoi l'utiliser

Fonctionnalités clés

90 tools across 11 categories for comprehensive browser automation
Anti-bot bypass for Cloudflare, Queue-It, and similar protections
Pixel-perfect UI element cloning with CDP-based DOM extraction
Dynamic network hooks — AI generates Python functions to intercept traffic
Core mode (22 tools) available when 90 tools is too many for your client

Démo en direct

Aperçu en pratique

stealth-browser-mcp.replay ▶ prêt

0/0

Installer

Choisissez votre client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "stealth-browser-mcp": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/vibheksoni/stealth-browser-mcp"
      ]
    }
  }
}

Ouvrez Claude Desktop → Settings → Developer → Edit Config. Redémarrez après avoir enregistré.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "stealth-browser-mcp": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/vibheksoni/stealth-browser-mcp"
      ]
    }
  }
}

Cursor utilise le même schéma mcpServers que Claude Desktop. La config projet l'emporte sur la globale.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "stealth-browser-mcp": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/vibheksoni/stealth-browser-mcp"
      ]
    }
  }
}

Cliquez sur l'icône MCP Servers dans la barre latérale Cline, puis "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "stealth-browser-mcp": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/vibheksoni/stealth-browser-mcp"
      ]
    }
  }
}

Même format que Claude Desktop. Redémarrez Windsurf pour appliquer.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "stealth-browser-mcp",
      "command": "TODO",
      "args": [
        "See README: https://github.com/vibheksoni/stealth-browser-mcp"
      ]
    }
  ]
}

Continue utilise un tableau d'objets serveur plutôt qu'une map.

~/.config/zed/settings.json

{
  "context_servers": {
    "stealth-browser-mcp": {
      "command": {
        "path": "TODO",
        "args": [
          "See README: https://github.com/vibheksoni/stealth-browser-mcp"
        ]
      }
    }
  }
}

Ajoutez dans context_servers. Zed recharge à chaud à la sauvegarde.

claude mcp add stealth-browser-mcp -- TODO 'See README: https://github.com/vibheksoni/stealth-browser-mcp'

Une seule ligne. Vérifiez avec claude mcp list. Supprimez avec claude mcp remove.

Cas d'usage

Usages concrets : stealth-browser-mcp

How to scrape data from a Cloudflare-protected site for authorized research

👤 Researchers and data analysts who need data from protected sites they have permission to access ⏱ ~15 min intermediate

Quand l'utiliser : You need to extract data from a site you're authorized to access but standard scraping tools get blocked.

Prérequis

stealth-browser-mcp installed — Clone repo, pip install -r requirements.txt
Authorization to access the target site — Ensure you have permission to scrape the site

Déroulement

Navigate to the site

Open a stealth browser and navigate to https://example.com/data. Wait for the page to fully load.✓ Copié

→ Page loaded successfully, bypassing protection
Extract data

Query all elements matching the product listing cards. Extract titles, prices, and URLs.✓ Copié

→ Structured data from the page

Résultat : Data extracted from the protected site for authorized analysis.

Pièges

Running too many concurrent browsers — Keep to 1-2 browser instances at a time to avoid detection patterns

Combiner avec : filesystem

Clone a UI element pixel-perfect for design reference

👤 Frontend developers who want to study and replicate a UI pattern ⏱ ~10 min intermediate

Quand l'utiliser : You see a UI component you like and want the complete HTML/CSS for reference.

Prérequis

stealth-browser-mcp running — Start the MCP server

Déroulement

Navigate and identify

Open https://example.com and take a screenshot. I want to clone the navigation header.✓ Copié

→ Screenshot with page rendered
Clone the element

Clone the nav header element completely — HTML structure, CSS styles, events, and animations. Save to a file.✓ Copié

→ Complete element clone saved with all styles and assets

Résultat : A self-contained HTML/CSS clone of the UI element for reference.

Pièges

Cloned elements may depend on external fonts or scripts — Use extract_element_assets to also capture referenced resources

Combiner avec : filesystem

Combinaisons

Associez-le à d'autres MCPs pour un effet X10

stealth-browser-mcp + filesystem

Save scraped data and cloned UI elements to organized files

Scrape all product data from the page and save it as JSON to ~/data/products.json.✓ Copié

Outils

Ce que ce MCP expose

Outil	Entrées	Quand appeler
spawn_browser	options?: obj	Launch a new stealth browser instance
navigate	url: str	Navigate to a URL
query_elements	selector: str	Find elements by CSS selector
clone_element_complete	selector: str	Pixel-perfect clone of a DOM element with all styles
create_dynamic_hook	hook_function: str	Create a Python function to intercept network traffic
take_screenshot	selector?: str	Capture a screenshot of the page or specific element
execute_script	script: str	Execute JavaScript in the page context

Coût et limites

Coût d'exécution

Quota d'API: N/A — runs locally
Tokens par appel: 200–2000 tokens per tool call
Monétaire: Free
Astuce: Use core mode (22 tools) if your client struggles with the full 90-tool set.

Sécurité

Permissions, secrets, portée

Stockage des identifiants : No credentials stored by the server. If accessing authenticated sites, manage cookies via get/set_cookie tools.

Sortie de données : Connects to whatever URLs you navigate to via Chrome

⚠ This tool is designed for authorized security research, CTF competitions, and defensive analysis only. Do not use it against systems you don't own or have written authorization to test.
Anti-bot bypass capabilities should only be used on sites you have authorization to access.
Dynamic hooks execute arbitrary Python code — review all hook functions before running.
Ensure your scraping complies with each site's Terms of Service and applicable laws.

Dépannage

Erreurs courantes et correctifs

Chrome not found

The server uses nodriver which manages its own Chrome. Ensure Chrome is installed or let nodriver download it.

Vérifier : which google-chrome || which chromium

Anti-bot detection still triggered

Try adding delays between actions, reduce concurrent requests, or use different browser profiles.

Vérifier : Check the page for CAPTCHA or block messages

Element clone incomplete

Some elements use shadow DOM or iframe isolation. Use execute_cdp_command for low-level access.

Vérifier : Inspect the element in DevTools to check for shadow roots

Alternatives

stealth-browser-mcp vs autres

Alternative	Quand l'utiliser	Compromis
playwright MCP	You don't need anti-bot bypass and want standard browser automation	Standard Playwright with no stealth features but broader community support

Plus

Ressources

📖 Lire le README officiel sur GitHub

🐙 Voir les issues ouvertes

🔍 Parcourir les 400+ serveurs MCP et Skills