🤖 Gausium OpenAPI MCP Server

🔧 A powerful MCP server bridging AI models with Gausium robots

Control and monitor Gausium cleaning robots through Claude, Cursor, and other AI assistants

🚀 Quick Start • 📖 Documentation • 🛠️ Installation • 🎯 Examples • 🇨🇳 中文文档

🌟 What is this?

This MCP (Model Control Protocol) server enables seamless interaction between AI models and Gausium cleaning robots through a standardized interface. Perfect for building intelligent automation workflows with Claude Code, Cursor, and other MCP-compatible AI tools.

🔗 Repository: https://github.com/cfrs2005/mcp-gs-robot

🎯 Key Benefits

• 🤖 AI-First Design: Built specifically for AI assistant integration

• 🔄 Real-time Control: Monitor and command robots instantly

• 📊 Rich Data Access: Get detailed status, maps, and task reports

• 🛡️ Secure: OAuth-based authentication with environment variables

• 🌐 Universal: Works with Claude, Cursor, and any MCP client

Related MCP server: ROS2 MCP Server

🏗️ Architecture

The server follows a layered architecture that separates concerns and promotes maintainability:

🔄 MCP Protocol Flow

The diagram below shows how AI models interact with Gausium robots through the MCP protocol:

✨ Features

🛠️ Core MCP Tools

🧠 Smart Routing Tools (Enhanced in v0.1.12)

🔧 Advanced Workflows

• 🎛️ Automated Task Execution: Complete workflows from status → task selection → execution

• 📈 Batch Operations: Handle multiple robots simultaneously

• 🗺️ Map Management: Upload, download, and manage robot maps

• 📊 Report Generation: Generate PNG maps from task reports

• 🏗️ Site-based Tasks: Advanced task creation with building/floor context

🤝 Supported Robot Lines

M-line Robots (Traditional Cleaning Robots)

• OMNIE (OMNIE series) - Multi-purpose cleaning robot

• Vacuum 40 (40 series) - Vacuum cleaning robot

• Scrubber 50 (50 series) - Floor scrubbing robot

• Scrubber 75 (75 series) - Heavy-duty floor scrubbing robot

S-line Robots (Advanced Smart Robots, including SW series)

• Phantas (S series) - Phantom intelligent cleaning robot

• BEETLE (SW series) - Beetle smart cleaning robot

📁 Project Structure

The project follows a structured layout optimized for MCP development:

🗂️ mcp-gs-robot/
├── 📦 src/gs_openapi/           # Main package
│   ├── 🔌 api/                  # Direct API integrations
│   │   ├── 🤖 robots.py         # Robot management APIs
│   │   └── 🗺️ maps.py           # Map management APIs
│   ├── 🔐 auth/                 # Authentication layer
│   │   └── 🎫 token_manager.py  # OAuth token lifecycle
│   ├── ⚙️ config.py             # Configuration management
│   ├── 🔧 core/                 # Core functionality
│   │   ├── 📡 client.py         # HTTP client wrapper
│   │   └── 🛣️ endpoints.py      # API endpoint definitions
│   ├── 🔌 mcp/                  # MCP server implementation
│   │   └── 🌉 gausium_mcp.py    # Main MCP bridge
│   └── 🔄 workflows/            # Automated workflows
│       └── 🎯 task_engine.py    # Task automation engine
├── 📚 docs/                     # Documentation
│   ├── 🖼️ images/               # Visual documentation
│   ├── 📖 apis.md              # API documentation
│   └── 🧪 TESTING_GUIDE.md     # Testing instructions
├── 🚀 main.py                  # Application entry point
└── 📋 pyproject.toml           # Package configuration

🔍 Key Components

🚀 Quick Start

📦 Installation

Option 1: Install from PyPI (Recommended)

pip install mcp-gs-robot

Option 2: Install from Source

# Clone repository
git clone https://github.com/cfrs2005/mcp-gs-robot.git
cd mcp-gs-robot

# Setup with uv (recommended)
uv venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
uv pip install -e .

🔧 Configuration

Set up your Gausium API credentials:

# Required environment variables
export GS_CLIENT_ID="your_client_id"
export GS_CLIENT_SECRET="your_client_secret" 
export GS_OPEN_ACCESS_KEY="your_access_key"

🔑 Get credentials from

🏃‍♂️ Running the Server

# Start MCP server (stdio mode)
python -m gs_openapi.main
# or if installed via pip:
mcp-gs-robot

✅ Server starts using stdio transport (perfect for Claude Code)

🔌 Claude Code Integration

Method 1: Automatic installation with environment setup

# Add MCP server with environment variables
claude mcp add mcp-gs-robot \
  --env GS_CLIENT_ID="your_client_id" \
  --env GS_CLIENT_SECRET="your_client_secret" \
  --env GS_OPEN_ACCESS_KEY="your_access_key"

Method 2: Manual configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "mcp-gs-robot": {
      "command": "mcp-gs-robot",
      "env": {
        "GS_CLIENT_ID": "your_client_id",
        "GS_CLIENT_SECRET": "your_client_secret", 
        "GS_OPEN_ACCESS_KEY": "your_access_key"
      }
    }
  }
}

Method 3: Using environment file

If you prefer to use a .env file:

# Set global environment variables
export GS_CLIENT_ID="your_client_id"
export GS_CLIENT_SECRET="your_client_secret"
export GS_OPEN_ACCESS_KEY="your_access_key"

# Simple MCP installation
claude mcp add mcp-gs-robot

💡 Note: This MCP server uses stdio transport (not SSE), which is perfect for Claude Code integration

🎯 Examples

📱 Claude Code Usage

# In Claude Code, you can now use natural language:

"List all my robots"
# → Calls mcp__mcp-gs-robot__list_robots

"Get status of robot GS101-0100-V1P-B001" 
# → Calls mcp__mcp-gs-robot__get_robot_status

"Start cleaning task for robot in building 5"
# → Orchestrates site info → map selection → task creation

🖥️ IDE Integration

Cursor Configuration:

Cherry Studio Configuration:

🐛 Debugging

Monitor server logs for troubleshooting:

📖 Documentation

🤝 Contributing

We welcome contributions! Please:

1. 🍴 Fork the repository

2. 🌿 Create a feature branch

3. ✅ Add tests for your changes

4. 📝 Update documentation

5. 🔄 Submit a pull request

📄 License

MIT License - see LICENSE file for details.

🆘 Support

• 📝 Issues

• 📧 Email

• 📚 Gausium Developer Docs

Made with ❤️ for the Claude Code community

Enabling AI-powered robot automation, one task at a time 🤖✨