arduino-mcp-server
Enables AI assistants to compile, upload, and monitor Arduino boards via natural language, with electrical safety checks and dependency management.
README
arduino-mcp-server
Give your AI assistant full control over Arduino — compile, upload, monitor serial, and verify wiring safety, all through natural language.
Part of the HardwareMCP ecosystem — open-source MCP servers that bridge AI to physical hardware.
What this does
AI assistants can control Jira, GitHub, and databases. They can't talk to a microcontroller — until now.
arduino-mcp-server wraps arduino-cli into an MCP server so your AI can:
- Detect connected boards and ports automatically
- Compile and upload sketches without touching the terminal
- Monitor serial output with stateful sessions (open, read, expect, write, close)
- Run electrical safety checks before sending commands to hardware
- Manage dependencies — cores, libraries, and CLI installation
Quick Start
Install:
npm install -g arduino-mcp-server
Add to Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"arduino": {
"command": "npx",
"args": ["-y", "arduino-mcp-server"],
"env": {
"ARDUINO_CLI_PATH": "arduino-cli",
"ARDUINO_SKETCH_ROOT": "/path/to/your/sketches"
}
}
}
}
Requires arduino-cli on your PATH, or let the server install it for you.
What you can say
Bootstrap from scratch:
"Check if Arduino CLI is installed and set everything up for an Arduino Uno."
Compile and upload:
"Compile my Blink sketch and upload it to the Uno on COM6."
Serial monitoring:
"Open serial on COM6 at 115200 and wait until the device prints READY."
Safety-first workflows:
"Run a safety preflight for an Arduino Uno with 5V on pin 13 at 25mA before I send commands."
Tools
| Tool | What it does |
|---|---|
arduino_cli_doctor |
Check Arduino CLI installation and version |
install_arduino_cli |
Guide through arduino-cli installation |
detect_hardware |
Detect connected boards and infer FQBNs |
list_connected_boards |
List all connected Arduino boards |
list_serial_ports |
List available serial ports |
ensure_core_installed |
Check/install board cores |
compile_sketch |
Compile a sketch for a target board |
upload_sketch |
Upload compiled sketch to a board |
upload_and_wait_ready |
Upload and wait for device ready signal |
serial_open_session |
Open a stateful serial session |
serial_read |
Read buffered serial data |
serial_expect |
Wait for a pattern in serial output |
serial_write |
Send data over serial |
serial_close_session |
Close a serial session |
serial_list_sessions |
List active serial sessions |
read_serial_snapshot |
Quick one-shot serial read |
safety_preflight |
Electrical safety check before hardware ops |
get_board_details |
Get pin/capability details for a board |
list_supported_boards |
List all boards arduino-cli supports |
list_board_reference |
Browse board pin reference |
search_board_reference |
Search board reference by keyword |
Resources:
arduino://boards/reference— structured board pin/capability reference
Prompts:
arduino-cli-bootstrap-policy— policy for arduino-cli setup behaviorarduino-setup-assistant— guided Arduino environment setup
Configuration
| Variable | Default | Description |
|---|---|---|
ARDUINO_CLI_PATH |
arduino-cli |
Path to arduino-cli binary |
ARDUINO_SKETCH_ROOT |
(none) | Restrict sketch paths to this directory |
Development
git clone https://github.com/hardware-mcp/arduino-mcp-server
cd arduino-mcp-server
npm install
npm run typecheck
npm run build
npm run dev
Part of HardwareMCP
This server is part of the HardwareMCP ecosystem — a collection of MCP servers that give AI assistants real control over physical hardware.
License
MIT — see LICENSE.
Support
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。