Skip to main content
Glama

rata-mcp

Official Model Context Protocol (MCP) server for RATA (GHM Automate v2) — Antidetect Browser Automation & AI Vision Engine.

Connect AI agents (Claude Desktop, Cursor, Antigravity, Cline...) directly to your antidetect browser profiles with Vision-based element interaction and automated workflows.

IMPORTANT

RATA is GHM Automate v2: GHM Automate (v1) provides the solid core automation foundation, while RATA (v2) elevates the ecosystem into an autonomous, AI-driven automation suite featuring Computer Vision (Set-of-Mark), live XPath reflection, and multi-profile agent control.

This package is a lightweight open-source MCP adapter. It requires the RATA / GHM Automate application (by BHIT TECH) running on your machine with its Gateway on port 1408.


🚀 Capabilities

  • System & Profiles: rata_health, rata_list_profiles, rata_get_job_status (with backward-compatible aliases for ghm_*).

  • RATA Browser Vision (rata_browser_*):

    • rata_browser_navigate: Open profile and navigate to URL (headed or headless).

    • rata_browser_get_state: Capture current page state with numbered interactive elements (Set-of-Mark) + screenshot.

    • rata_browser_click, rata_browser_type, rata_browser_scroll, rata_browser_press_key, rata_browser_upload_file.

    • rata_browser_evaluate_js: Run inspected scripts and test resilient XPath locators live.

    • rata_browser_switch_tab, rata_browser_close_tab, rata_browser_close_session.

  • Dynamic Script Runners: Auto-discovered high-level automation runners exposed by the local application (Facebook Fanpage, Groups, TikTok, etc.).


Related MCP server: MCP Browser

📦 Quick Start

1. Requirements

  • Node.js >= 18

  • RATA (GHM Automate v2) desktop app running with Gateway enabled (port 1408).

2. Configuration for AI Clients

Antigravity / Claude Desktop / Cursor

Add to your MCP configuration:

{
  "mcpServers": {
    "rata": {
      "command": "npx",
      "args": [
        "-y",
        "rata-mcp"
      ],
      "env": {
        "RATA_GATEWAY_URL": "http://127.0.0.1:1408"
      }
    }
  }
}

Or connect via local build:

{
  "mcpServers": {
    "rata": {
      "command": "node",
      "args": [
        "path/to/ghmautomate/packages/mcp-server/dist/index.js"
      ]
    }
  }
}

⚙️ Environment Variables

Variable

Default

Description

RATA_GATEWAY_URL

http://127.0.0.1:1408

URL of the local RATA Gateway

GHM_GATEWAY_URL

http://127.0.0.1:1408

Backward compatibility fallback URL

RATA_GATEWAY_TOKEN

(none)

Bearer authentication token if enabled on Gateway

GHM_GATEWAY_TOKEN

(none)

Backward compatibility fallback token


🔒 Security

  • Keep the Gateway bound to 127.0.0.1; never expose port 1408 to the public network without proper VPN/tunneling and token authorization.

  • rata_browser_evaluate_js executes JavaScript in the context of the active profile. Only connect trusted AI clients.


🛠️ Build from Source

npm install
npm run build
node dist/index.js

📄 License

MIT License for this MCP adapter. The core RATA / GHM Automate desktop application is proprietary commercial software of BHIT TECH and is not covered by this license.

Maintenance

ActivityMaintained
ResponsivenessNo issues

Related MCP Connectors

Related MCP Servers

  • A
    license
    A
    quality
    B
    maintenance
    MCP server for browser automation with anti-detection. Scout pages, find elements, interact with websites, and monitor network traffic from any AI client that supports the Model Context Protocol.
    21
    1
    MIT
  • A
    license
    A
    quality
    D
    maintenance
    MCP server that lets AI assistants browse the web through your real Chrome with your cookies, sessions, and fingerprint, avoiding bot detection and CAPTCHAs. Enables web browsing, content extraction, and multi-step workflows via persistent tabs.
    15
    8 npm
    3
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    An MCP server that exposes browser-drive tools over a local endpoint, letting AI agents launch and control isolated browser profiles with their own cookies, fingerprints, and proxies, then navigate, click, type, extract, screenshot, evaluate JS, and manage tabs and cookies. Session state persists across launches so authenticated workflows carry over between AI sessions.
    MIT