by vibheksoni
The only browser automation that bypasses anti-bot systems. AI writes network hooks, clones UIs pixel-perfect via simple chat.
# Add to your Claude Code skills
git clone https://github.com/vibheksoni/stealth-browser-mcpGuides for using ai agents skills like stealth-browser-mcp.
Last scanned: 5/10/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-05-10T06:36:31.328Z",
"semgrepRan": false,
"npmAuditRan": true,
"pipAuditRan": false
}stealth-browser-mcp is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by vibheksoni. The only browser automation that bypasses anti-bot systems. AI writes network hooks, clones UIs pixel-perfect via simple chat. It has 1,602 GitHub stars.
Yes. stealth-browser-mcp passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/vibheksoni/stealth-browser-mcp" and add it to your Claude Code skills directory (see the Installation section above).
stealth-browser-mcp is primarily written in Python. It is open-source under vibheksoni on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh stealth-browser-mcp against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
Stealth browser automation for MCP-compatible AI agents.
Navigate Cloudflare challenges, anti-bot checks, and login walls with real Chrome-family browser instances powered by nodriver, Chrome DevTools Protocol, and FastMCP.
NodeMaven is a proxy provider offering high-quality IPs for browser automation, web scraping, SEO research, and social media management.
Why NodeMaven?
Special discounts for stealth-browser-mcp users
STEALTHMCP35 — 35% off Mobile and Residential ProxiesSTEALTHMCP40 — 40% off ISP Static ProxiesSponsored placement. NodeMaven is a paid sponsor of this project.
Stealth Browser MCP passing a Cloudflare challenge, cloning UI elements, and inspecting network traffic through AI chat commands.
Requires Python 3.10+ and Chrome, Chromium, or Microsoft Edge.
git clone https://github.com/vibheksoni/stealth-browser-mcp.git
cd stealth-browser-mcp
python -m venv venv
# Windows
venv\Scripts\activate
# macOS / Linux
source venv/bin/activate
pip install -r requirements.txt
Claude Code CLI (recommended):
Windows:
claude mcp add-json stealth-browser-mcp "{\"type\":\"stdio\",\"command\":\"C:\\path\\to\\stealth-browser-mcp\\venv\\Scripts\\python.exe\",\"args\":[\"C:\\path\\to\\stealth-browser-mcp\\src\\server.py\"]}"
macOS / Linux:
claude mcp add-json stealth-browser-mcp '{
"type": "stdio",
"command": "/path/to/stealth-browser-mcp/venv/bin/python",
"args": ["/path/to/stealth-browser-mcp/src/server.py"]
}'
Replace the example path with your real clone path.
Windows (%APPDATA%\Claude\claude_desktop_config.json):
{
"mcpServers": {
"stealth-browser-mcp": {
"command": "C:\\path\\to\\stealth-browser-mcp\\venv\\Scripts\\python.exe",
"args": ["C:\\path\\to\\stealth-browser-mcp\\src\\server.py"],
"env": {}
}
}
}
macOS / Linux (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"stealth-browser-mcp": {
"command": "/path/to/stealth-browser-mcp/venv/bin/python",
"args": ["/path/to/stealth-browser-mcp/src/server.py"],
"env": {}
}
}
}
These commands are provided for users who prefer FastMCP's installer. The stdio JSON setup above is the recommended path.
pip install fastmcp
fastmcp install claude-desktop src/server.py --with-requirements requirements.txt
fastmcp install claude-code src/server.py --with-requirements requirements.txt
fastmcp install cursor src/server.py --with-requirements requirements.txt
Restart your MCP client and ask your agent:
Use stealth-browser to navigate to https://example.com and take a screenshot.
This repo includes an agent skill at skills/stealth-browser-mcp for clients that support Codex-style skills. It teaches agents the recommended tool order, state checks, pre-document CDP script usage, network debugging flow, and browser cleanup rules.
Use the skill when an LLM struggles to choose the right MCP tool or forgets to verify browser state after actions. If your client does not load repo-local skills automatically, add or symlink skills/stealth-browser-mcp into your client skills directory.
This server exposes powerful local browser-control primitives: CDP execution, page JavaScript execution, cookie/storage access, network interception, dynamic hooks, header mutation, navigation, and file uploads from allowlisted local paths.
Treat the MCP client or agent as the security principal. The recommended deployment is local stdio transport with a trusted desktop MCP client. If you use HTTP transport, set STEALTH_BROWSER_MCP_AUTH_TOKEN and do not expose unauthenticated HTTP outside trusted local or private networks.
Choose exactly what functionality you need. Run the full 97-tool suite or strip it down to the 20-tool core.
| Mode | Tools | Use Case |
|---|---|---|
| Full (default) | 97 | Complete browser automation and debugging |
Minimal (--minimal) |
20 | Core browser automation and interaction |
Custom (--disable-*) |
Your choice | Disable specific sections |
Xpool safe (--xpool-safe) |
83 | Disable CDP function tools that trigger Runtime.enable |
python src/server.py --minimal
python src/server.py --disable-cdp-functions --disable-dynamic-hooks
python src/server.py --list-sections
python src/server.py --debug
python src/server.py --transport http --host 127.0.0.1 --port 8000
Use --debug or set STEALTH_BROWSER_DEBUG=1 to enable verbose server diagnostics on stderr. Normal MCP stdio runs stay quiet by default to avoid corrupting JSON-RPC transport output.
Available sections:
| Section | Tools | Description |
|---|---|---|
browser-management |
8 | Core browser operations |
element-interaction |
12 | Page interaction and manipulation |
element-extraction |
9 | Element cloning and extraction |
file-extraction |
9 | File-based extraction tools |
network-debugging |
10 | Network monitoring and capture |
cdp-functions |
14 | Chrome DevTools Protocol execution |
progressive-cloning |
10 | Advanced element cloning |
cookies-storage |
3 | Cookie management |
tabs |
5 | Tab management |
debugging |
7 | Debug and system tools |
dynamic-hooks |
10 | AI-assisted network hooks |
These are environment variables for the MCP server process. Set them in your shell, service manager, Docker environment, or MCP client env block.
| Variable | Default | Meaning |
|---|---|---|
STEALTH_BROWSER_MCP_AUTH_TOKEN |
unset | Optional bearer token for HTTP transport. When set, HTTP clients must send Authorization: Bearer <token>. |
MCP_AUTH_TOKEN |
unset | Backward-compatible alias for STEALTH_BROWSER_MCP_AUTH_TOKEN. |
BROWSER_IDLE_TIMEOUT |
600 |
Global idle timeout in seconds before an unused browser instance is auto-clo |