mcp-arduino-server
MCP Arduino Server is a bridge that exposes Arduino CLI functionalities over the Model Context Protocol, enabling users to manage and deploy Arduino projects efficiently. It leverages AI to generate schematics and supports integration with MCP clients for streamlined operations.
MCP Arduino Server (mcp-arduino-server)
A FastMCP-powered bridge exposing arduino-cli
functionality via the Model Context Protocol (MCP). Manage sketches, boards, libraries, files, plus generate WireViz schematics from YAML or natural language.
Requirements
- Python ≥3.10
- arduino-cli in
PATH
- MCP SDK (
mcp[cli]
) - WireViz (optional; for diagram generation)
- OPENAI_API_KEY (for AI‑powered WireViz)
- thefuzz[speedup] (optional; enables fuzzy local library search)
Installation
From PyPI:
pip install mcp-arduino-server
From source:
git clone https://github.com/Volt23/mcp-arduino-server.git
cd mcp-arduino-server
pip install .
Configuration
Environment variables override defaults:
Variable | Default / Description |
---|---|
ARDUINO_CLI_PATH | auto-detected |
WIREVIZ_PATH | auto-detected |
MCP_SKETCH_DIR | ~/Documents/Arduino_MCP_Sketches/ |
LOG_LEVEL | INFO |
OPENAI_API_KEY | your OpenAI API key (required for AI‑powered WireViz) |
OPENROUTER_API_KEY | optional alternative to OPENAI_API_KEY |
Quick Start
mcp-arduino-server
Server listens on STDIO for JSON-RPC MCP calls. Key methods:
Sketches
create_new_sketch(name)
list_sketches()
read_file(path)
write_file(path, content[, board_fqbn])
(auto-compiles & opens.ino
)
Build & Deploy
verify_code(sketch, board_fqbn)
upload_sketch(sketch, port, board_fqbn)
Libraries
lib_search(name[, limit])
lib_install(name)
list_library_examples(name)
Boards
list_boards()
board_search(query)
File Ops
rename_file(src, dest)
remove_file(path)
(destructive; operations sandboxed to home & sketch directories)
WireViz Diagrams
generate_circuit_diagram_from_description(desc, sketch="", output_base="circuit")
(AI‑powered; requiresOPENAI_API_KEY
, opens PNG automatically)
MCP Client Configuration
To integrate with MCP clients (e.g., Claude Desktop), set your OpenAI API key in the environment (or alternatively OPENROUTER_API_KEY
for OpenRouter):
{
"mcpServers": {
"arduino": {
"command": "/path/to/mcp-arduino-server",
"args": [],
"env": {
"WIREVIZ_PATH": "/path/to/wireviz",
"OPENAI_API_KEY": "<your-openai-api-key>"
}
}
}
}
Troubleshooting
- Set
LOG_LEVEL=DEBUG
for verbose logs. - Verify file and serial-port permissions.
- Install missing cores:
arduino-cli core install <spec>
. - Run
arduino-cli
commands manually to debug.
License
MIT