Skip to main content
Glama
sandraschi

Robotics MCP Server

by sandraschi
OverviewSchemaRelated ServersScoreDiscussions
Python
Hybrid

Robotics MCP Server

By FlowEngineer sandraschi

Unified robotics control via MCP - Physical and virtual robots (bot + vbot)

⚠️ CRITICAL REQUIREMENTS

Hardware (Recommended)

  • Physical Robot: Moorebot Scout, Unitree Go2/G1/H1

  • Without hardware: Virtual robotics only (Unity3D + VRChat)

Software (MANDATORY)

Without these, the system will not function. See Complete Setup Guide.

FastMCP Python Unity VRChat License Status

⚠️ REQUIRED: Prerequisites & Dependencies

🔴 PHYSICAL ROBOTS (RECOMMENDED)

You SHOULD own one of these supported robots:

  • Moorebot Scout (primary focus) - Mecanum wheeled robot with LiDAR

  • Unitree Go2 - Advanced quadrupedal robot

  • Unitree G1 - Humanoid robot with arms

  • Unitree H1 - Humanoid robot platform

Physical robots provide the complete robotics experience. Virtual robots are for testing/simulation.

🟡 REQUIRED SOFTWARE

You MUST install these applications:

Unity 3D (Required for virtual robotics)

# Download and install Unity Hub from: # https://unity.com/download # Then install Unity Editor version 6000.2.14f1 or later: # 1. Open Unity Hub # 2. Go to "Installs" tab # 3. Click "Add" → "Official releases" # 4. Select "6000.2.14f1" (LTS recommended) # 5. Install with default components + Android Build Support

VRChat (Required for social VR robotics)

# Download from Steam: # https://store.steampowered.com/app/438100/VRChat/ # Or from VRChat website: # https://hello.vrchat.com/

🟢 REQUIRED MCP SERVERS

You MUST install and configure these MCP servers:

1. Unity3D-MCP (Virtual robot control)

# Clone and install: git clone https://github.com/sandraschi/unity3d-mcp.git cd unity3d-mcp pip install -e . # Add to Cursor MCP configuration

2. OSC-MCP (Real-time communication)

# Clone and install: git clone https://github.com/sandraschi/osc-mcp.git cd osc-mcp pip install -e . # Add to Cursor MCP configuration

3. VRChat-MCP (Social VR integration)

# Clone and install: git clone https://github.com/sandraschi/vrchat-mcp.git cd vrchat-mcp pip install -e . # Add to Cursor MCP configuration

4. Blender-MCP (3D model creation)

# Clone and install: git clone https://github.com/sandraschi/blender-mcp.git cd blender-mcp pip install -e . # Requires Blender 4.0+ installed

5. Avatar-MCP (Avatar management)

# Clone and install: git clone https://github.com/sandraschi/avatar-mcp.git cd avatar-mcp pip install -e .

⚠️ Important: Alpha Status & Current State

This server is in ALPHA status with Cursor MCP integration working:

  • ✅ Cursor MCP Integration: Server now starts successfully in Cursor IDE

  • ✅ Unity Integration: Now enabled with robust error handling and fallbacks

  • ⏸️ Other MCP Servers: Composition temporarily limited for stability

  • Status: Active development - features may change, break, or be incomplete

  • Virtual Robotics: Prioritized (vbot) - physical robot support coming after hardware arrives (XMas 2025)

  • Core Functionality: 7 portmanteau tools available for robot control and management

See Cursor MCP Setup section for integration details.

🎯 Overview

Robotics MCP Server provides unified control for both physical robots (ROS-based) and virtual robots (Unity/VRChat), with a focus on Moorebot Scout, Unitree robots, and virtual robotics testing.

🚀 Project Stats: ~9,200 lines of code, ~2,600 lines of tests, ~4,100 lines of documentation

⚠️ ALPHA STATUS - CURSOR INTEGRATION WORKING: Server successfully starts in Cursor MCP mode. Mounted servers temporarily disabled for stability.

Key Features

  • Physical Robot Control: Moorebot Scout (ROS 1.4), Unitree Go2/G1

  • YDLIDAR SuperLight (95g) LiDAR integration for Scout

  • Virtual Robot Control: Unity3D/VRChat/Resonite integration via existing MCP servers

  • ROS Bridge Integration: Robot Operating System 1.4 (Melodic) via rosbridge_suite

  • LiDAR Sensing: Affordable 3D LiDAR Guide - Livox Mid-360 ($399), RPLIDAR ($99), and more

  • Tiny Controllers: Pico & Micro Boards - Raspberry Pi Pico, ESP32, Arduino Nano for small robots

  • Motion Detection: Pyroelectric Sensors Guide - AM312, HC-SR501 ultra-small PIR sensors ($1-5)

  • Multi-Robot Coordination: Physical and virtual robots together

  • World Labs Marble/Chisel: Environment generation and import

  • Dual Transport: stdio (MCP) + HTTP (FastAPI) endpoints

  • MCP Server Composition: Ready for integration with osc-mcp, unity3d-mcp, vrchat-mcp, avatar-mcp, blender-mcp, gimp-mcp (temporarily disabled)

  • 7 Portmanteau Tools: robotics_system, robot_control, robot_behavior, robot_manufacturing, robot_virtual, robot_model_tools, vbot_crud

  • Robot Model Creation: Framework ready for automated 3D model creation

📚 Documentation

🚀 Quick Start

Prerequisites Check

⚠️ BEFORE STARTING: Complete all Setup Prerequisites - Unity3D, VRChat, and MCP servers are REQUIRED.

Cursor MCP Integration

✅ The robotics-mcp server now works in Cursor IDE!

Setup Steps:

  1. Complete Prerequisites: Install Unity3D, VRChat, and all required MCP servers

  2. Install the package: pip install -e ".[dev]"

  3. Add to Cursor MCP configuration using the provided mcpb.json

  4. The server will automatically start when enabled in Cursor

Available Tools:

  • robotics_system - System management (help, status, list_robots)

  • robot_control - Physical robot control (ready for hardware)

  • robot_behavior - Robot behavior and animation

  • robot_manufacturing - 3D printing and CNC control

  • robot_virtual - Virtual robotics operations

  • robot_model_tools - Model creation and conversion

  • vbot_crud - Virtual robot lifecycle management

MCP Server Integration

✅ ENABLED WITH SAFETY: Unity3D-MCP is now enabled with robust error handling, timeouts, and fallbacks.

Active Integration:

  • osc-mcp: ✅ ENABLED - OSC communication for real-time robot control

  • unity3d-mcp: ✅ ENABLED - Unity3D integration for virtual robotics

  • vrchat-mcp: ⏸️ DISABLED - VRChat integration (protocol conflicts)

  • avatar-mcp: ⏸️ DISABLED - Avatar management (timeseries conflicts)

  • blender-mcp: ⏸️ DISABLED - 3D model creation (protocol hangs)

  • gimp-mcp: ⏸️ DISABLED - Texture creation (protocol hangs)

Safety Features:

  • 30-second timeouts for server loading

  • 3 retry attempts with exponential backoff

  • Graceful fallbacks to mock operations if Unity unavailable

  • Never blocks robotics-mcp server operation

  • Comprehensive logging for debugging

Configuration: All required MCP servers are automatically loaded with error protection. If Unity is not available, virtual robot operations fall back to mock mode with full functionality preservation.

Configuration (Optional)

The server works out-of-the-box without configuration. For advanced setups, create ~/.robotics-mcp/config.yaml:

robotics: moorebot_scout: enabled: false robot_id: "scout_01" ip_address: "192.168.1.100" port: 9090 mock_mode: true virtual: enabled: true platform: "unity" server: enable_http: true http_port: 8080 log_level: "INFO"

MCP Integration Config (for when mounted servers are re-enabled):

mcp_integration: osc_mcp: enabled: true prefix: "osc" unity3d_mcp: enabled: true prefix: "unity" vrchat_mcp: enabled: true prefix: "vrchat" avatar_mcp: enabled: true prefix: "avatar" blender_mcp: enabled: true prefix: "blender" gimp_mcp: enabled: true prefix: "gimp"

Running the Server

Primary Usage: Configure as MCP server in Cursor IDE using mcpb.json

Manual Operation (Development)

# MCP stdio mode (for testing) python -m robotics_mcp --mode stdio # HTTP API mode python -m robotics_mcp --mode http --port 8080 # Dual mode (stdio + HTTP) python -m robotics_mcp --mode dual --port 8080

🛠️ Usage

MCP Tools

Robot Control

# Get robot status await robot_control(robot_id="scout_01", action="get_status") # Move robot await robot_control( robot_id="scout_01", action="move", linear=0.2, angular=0.0 ) # Stop robot await robot_control(robot_id="scout_01", action="stop")

Virtual Robotics

# Spawn virtual robot in Unity await virtual_robotics( robot_type="scout", action="spawn_robot", platform="unity", position={"x": 0, "y": 0, "z": 0} ) # Load Marble environment await virtual_robotics( action="load_environment", environment="stroheckgasse_apartment", platform="unity" )

Robot Model Tools

# Create Scout model from scratch (uses blender-mcp + gimp-mcp) await robot_model_create( robot_type="scout", output_path="D:/Models/scout_model.fbx", format="fbx", dimensions={"length": 0.115, "width": 0.10, "height": 0.08}, create_textures=True, texture_style="realistic" ) # Import robot model into Unity await robot_model_import( robot_type="scout", model_path="D:/Models/scout_model.fbx", format="fbx", platform="unity", project_path="D:/Projects/UnityRobots" ) # Convert model between formats await robot_model_convert( source_path="D:/Models/scout.fbx", source_format="fbx", target_format="glb", target_path="D:/Models/scout.glb" )

HTTP API

Health Check

curl http://localhost:8080/api/v1/health

List Robots

curl http://localhost:8080/api/v1/robots

Control Robot

curl -X POST http://localhost:8080/api/v1/robots/scout_01/control \ -H "Content-Type: application/json" \ -d '{"action": "move", "linear": 0.2, "angular": 0.0}'

List Tools

curl http://localhost:8080/api/v1/tools

Call Tool

curl -X POST http://localhost:8080/api/v1/tools/robot_control \ -H "Content-Type: application/json" \ -d '{"robot_id": "scout_01", "action": "get_status"}'

📚 Documentation

🧪 Testing

Comprehensive test suite: 21 test files, 2,642 lines of tests covering all 11 tools!

# Run all tests pytest # Run unit tests only pytest tests/unit # Run integration tests pytest tests/integration # Run with coverage pytest --cov=robotics_mcp --cov-report=html # Or use the PowerShell script .\scripts\run-tests.ps1

🔧 Development

Project Structure

robotics-mcp/ ├── src/robotics_mcp/ │ ├── server.py # Main FastMCP server │ ├── clients/ # Robot client implementations │ ├── integrations/ # MCP server integration wrappers │ ├── tools/ # Portmanteau tool implementations │ └── utils/ # Utilities (config, state, mock data) ├── tests/ │ ├── unit/ # Unit tests │ └── integration/ # Integration tests ├── docs/ # Documentation ├── scripts/ # Utility scripts └── mcpb/ # MCPB packaging

Code Quality

# Format code black src/ tests/ # Lint code ruff check src/ tests/ # Type checking mypy src/

🔧 Troubleshooting

Cursor MCP Integration Issues

Problem: Server not appearing in Cursor MCP tools

Symptoms:

  • "robotics-mcp" not showing in MCP tools list

  • Tools not available in Cursor

Quick Fix:

  1. Open Cursor Settings → Features → Model Context Protocol

  2. Add new server using the mcpb.json configuration

  3. Ensure the server shows as "Healthy" in the list

  4. Restart Cursor if needed

Server Status Check:

# Verify server starts correctly python -c "from robotics_mcp.server import RoboticsMCP; RoboticsMCP(); print('SUCCESS')"

Configuration File: Use mcpb.json in the project root for Cursor MCP setup.

Other Issues

Server won't start:

  • Check Python version: python --version (requires 3.10+)

  • Verify dependencies: pip install -e ".[dev]"

  • Check logs: C:\Users\sandr\AppData\Roaming\Cursor\logs\

Tools not appearing:

  • Verify MCP server is enabled in Cursor IDE settings

  • Check server logs for errors

  • Try disabling and re-enabling the server

Unity integration not working:

  • Ensure Unity Editor is running

  • Verify Unity project path is correct

  • Check unity3d-mcp server is healthy

🤝 Contributing

Contributions welcome! Please see CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE for details.

🙏 Acknowledgments

  • FastMCP framework

  • ROS community

  • Unity3D, VRChat, World Labs Marble/Chisel

  • MCP ecosystem contributors

Status: ✅ ALPHA - Cursor Integration Working - Core robotics-mcp functionality operational, mounted servers temporarily disabled for stability

⏸️ MCP Server Composition: Currently runs standalone. Integration with osc-mcp, unity3d-mcp, vrchat-mcp, avatar-mcp, blender-mcp, gimp-mcp planned for future releases.

Install Server
A
security – no known vulnerabilities
A
license - permissive license
A
quality - confirmed to work

How are these scores calculated?

Resources

Tools

New MCP Servers

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/sandraschi/robotics-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server