Buddy MCP Server
MCP server for controlling compatible boards over Bluetooth serial, including DC motors, servos, IO, and sensors.
README
Buddy MCP Server
English | 한국어
MCP server for compatible board over Bluetooth serial (DC motors, servos, IO, sensors).
Requires Python 3.10+, compatible board + Bluetooth dongle.
Run from a git clone using the project’s virtual environment. Clone this repo and run via run.py (uses .venv, created automatically if missing). Do not use global Python or other envs. This server is intended for local use only (config file + stdio).
Local setup (step-by-step) — Windows, macOS, Linux
Use the same steps on all platforms. Only the config file path and, on Windows, the Python command differ.
Step 1 — Clone and enter the project
git clone https://github.com/johnsnow-nam/buddy-mcp.git buddy-mcp
cd buddy-mcp
On Windows you can use Command Prompt, PowerShell, or Git Bash.
Step 2 — Print MCP config JSON
Run this in the project directory. The output is the JSON to add to your config file.
| OS | Command |
|---|---|
| macOS / Linux | python3 run.py --print-mcp-config |
| Windows | python run.py --print-mcp-config or py run.py --print-mcp-config |
Copy the entire JSON output (it includes the correct path to run.py for your machine). Example shape:
{
"mcpServers": {
"buddy": {
"command": "python3",
"args": ["/absolute/path/to/buddy-mcp/run.py"]
}
}
}
Step 3 — Config file location
Open (or create) the config file at the path for your OS:
| OS | Config file path |
|---|---|
| macOS | ~/Library/Application Support/Claude/claude_desktop_config.json |
| Windows | %APPDATA%\Claude\claude_desktop_config.json (e.g. C:\Users\<YourName>\AppData\Roaming\Claude\claude_desktop_config.json) |
| Linux | ~/.config/Claude/claude_desktop_config.json |
- To open on macOS:
open "$HOME/Library/Application Support/Claude/claude_desktop_config.json"(quote the path because of the space). - To open on Windows: run
%APPDATA%\Claude\claude_desktop_config.jsonin Explorer address bar or use Notepad/VS Code. - To open on Linux: e.g.
xdg-open ~/.config/Claude/claude_desktop_config.jsonor edit with your editor.
Step 4 — Add Buddy MCP to the config file
- If the file is empty or does not exist: paste the full JSON from Step 2 as the file content.
- If the file already has an
mcpServerssection: merge the"buddy"block from Step 2 into that section. Keep valid JSON (no trailing commas). Example with another server:
{
"mcpServers": {
"other-server": { ... },
"buddy": {
"command": "python3",
"args": ["/path/to/buddy-mcp/run.py"]
}
}
}
Save the file.
Step 5 — Restart the client
Restart Claude Desktop (or Cursor, if you use .cursor/mcp.json in the project instead). The client will start the Buddy MCP server locally via run.py.
Step 6 — Use the tools
In chat:
- Run buddy_list_ports to see serial ports.
- Run buddy_connect(port) with the port you want (e.g. the nRF or COM port). If you leave the port empty or use
"auto", a port whose description starts with “nRF” is chosen automatically when available. - Then use buddy_send_dc, buddy_send_servo, buddy_send_io, buddy_sensor_config, buddy_disconnect as needed.
No BUDDY_PORT environment variable is required when using buddy_connect.
Connection mode (local)
The server runs in stdio mode: the client (Claude Desktop, Cursor, etc.) starts run.py as a subprocess and talks over stdin/stdout. Do not redirect stdout (e.g. 2>&1)—the client expects only JSON-RPC on stdout.
MCP setup — stdio (Cursor, etc.)
- Clone the repo:
git clone https://github.com/johnsnow-nam/buddy-mcp.git buddy-mcp cd buddy-mcp - Print MCP config (uses project
.venv; creates it if missing). The output contains the correct path torun.pyfor your machine:
Ifpython3 run.py --print-mcp-config./run.pygives "permission denied", usepython3 run.pyinstead (or runchmod +x run.pyonce). - Paste the printed JSON into your MCP config (e.g. Cursor
.cursor/mcp.json, or your client’smcpServers). Use the output of step 2 as-is; do not use the example path below. Do not add2>&1or other stdout redirection—the client expects only JSON-RPC on stdout.
MCP config example (for reference only; use the JSON from step 2):
{
"mcpServers": {
"buddy": {
"command": "python3",
"args": ["/path/to/buddy-mcp/run.py"]
}
}
}
MCP starts the server via run.py; run.py ensures the server runs inside the project’s .venv only (no global Python).
Tools
| Tool | Purpose |
|---|---|
buddy_list_ports |
List COM/serial ports |
buddy_connect(port) |
Connect to board (pick port from list) |
buddy_disconnect |
Disconnect |
buddy_send_dc, buddy_send_servo, buddy_send_io, buddy_sensor_config |
Control motors, IO, sensors |
No BUDDY_PORT needed: in chat, run buddy_list_ports → buddy_connect(port).
한국어: README.ko.md
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。