ardupilot -sandbox
ArduPilotドローンをAIエージェントから操作するMCPサーバーです。
claude mcp add --transport stdio hfujikawa77-ardupilot-mcp-server-sandbox python ardupilot_mcp_server.py
How to use
This MCP server lets you control an ArduPilot drone via natural language by connecting to a MAVLink-enabled SITL or real aircraft over TCP. The server exposes a set of MCP tools such as arming/disarming the vehicle, takeoff to a specified altitude, and changing flight modes. These tools translate natural language prompts into ArduPilot commands that are executed through the MAVLink link at tcp:127.0.0.1:5762. You can access and invoke this MCP server from MCP hosts like Claude Desktop or Cline, enabling seamless, chat-based drone operations.
To use it, ensure ArduPilot is running and MAVLink is reachable at the expected TCP endpoint. Then start the MCP server with the configured command. From your MCP host, issue commands such as “arm the vehicle”, “take off to 10 meters”, or “change mode to GUIDED” and the server will relay these requests to ArduPilot via MAVLink, returning status updates and results in natural language.
How to install
Prerequisites:
- Python 3.10+ installed on your system
- ArduPilot SITL or actual vehicle connected and MAVLink TCP endpoint available at tcp:127.0.0.1:5762
Step-by-step installation:
- Clone or download the MCP server repository for ArduPilot Sandbox.
- Create and activate a Python virtual environment (recommended):
- python -m venv venv
- source venv/bin/activate # Linux/macOS
- venv\Scripts\activate # Windows
- Install required Python packages:
- pip install -r requirements.txt
- Ensure ArduPilot SITL or real vehicle is running and MAVLink is listening on tcp:127.0.0.1:5762.
- Start the MCP server:
- python ardupilot_mcp_server.py
- Verify that the server announces it is ready to accept connections (e.g., readiness message in console).
Optional: If your environment requires adjustments to PYTHONPATH or additional dependencies, set them prior to launching the server.
Additional notes
Environment and connectivity tips:
- Ensure MAVLink TCP endpoint is accessible at tcp:127.0.0.1:5762; adjust accordingly if using a different host/port.
- ArduPilot SITL must be running and ready before starting the MCP server.
- The MCP server exposes tools like arm(), disarm(), takeoff(altitude=...), and change_mode(mode=...). These commands are mapped to ArduPilot actions; ensure safety procedures are followed when testing.
- If using Cline or Claude Desktop, configure the MCP server in your settings to point to ardupilot_mcp_server.py and pass any required environment setup (e.g., PYTHONPATH) as described in the README.
- Check firewall and network settings if the MCP host and ArduPilot bridge are on different machines.
Related MCP Servers
supermcp
🚀 SuperMCP - Create multiple isolated MCP servers using a single connector. Build powerful Model Context Protocol integrations for databases (PostgreSQL, MSSQL) with FastAPI backend, React dashboard, and token-based auth. Perfect for multi-tenant apps and AI assistants.
TexMCP
A small FastMCP-based Microservice that renders LaTeX to PDF. The server exposes MCP tools to render raw LaTeX or templates and produces artifacts
simple
A python implementation of the Model Context Protocol (MCP) server with fastmcp, fastapi and streamablehttp.
fastmcp-builder
A comprehensive Claude Code skill for building production-ready MCP servers using FastMCP. Includes reference guides, runnable examples, and a complete implementation with OAuth, testing, and best practices.
REI3-Tickets
A simple FastMCP server for the REI3 Tickets application. Designed to supercharge ticket management with AI!
math -learning
Educational MCP server with math operations, matrix algebra, data visualization, and persistent workspace using FastMCP 3.0