umeiko/umeko_serial_mcp

This page was translated automatically. Technical details may differ from the original. View the original

65 stars 11 forks Quality 81 Trend 22MIT

A great tool for microcontroller development using AI, allowing AI assistants to communicate directly with your microcontroller or embedded devices via serial via natural language.

Overview

Local serial communication service based on the MCP protocol, with a built-in real-time web monitoring panel. Enable AI assistants to communicate directly with your microcontroller and embedded devices through natural language. - 🖥️ — Seamless integration with AI clients via Model Context Protocol — 🔌 Automatic scanning, connection, and read/write of serial devices — 🌐 — Built-in HTTP + WebSocket dual-port service, real-time browser bypass monitoring - This project has been packaged and uploaded to the UV and PyPi repositories. Major AI assistants can directly add it via the following command line, and it will automatically pull and configure it. Kimi Code Configure the MCP server through mcp.json. Edit the user-level configuration ~/.kimi-code/mcp.json (or item-level .kimi-code/mcp.json), and add the following: You can also run /mcp-config in Kimi Code's TUI to interactively add it; After configuration, run /mcp to view the connection status. This is the default and recommended configuration format for Codex CLI. Open your configuration file (such as ~/.codex/config.toml) and add the following: 🦞Other clients such as OpenClaw, Cursor, Cline, TRAE, Cherry Studio, Qwen Chat, etc. In the MCP server settings, select to add and paste the following fields using JSON. ⚠️ ：uvx --from . It caches the wheel package. After modifying the source code, you need to first upgrade the pyproject.toml version, then run uv build --wheel, and finally restart the MCP client for it to take effect. If you want clients to use your locally deployed service, add the following in the mcp.json of MCP-supported clients (such as Claude-code, Cursor, Cline, etc.):

Overview (machine translated)

Local serial communication service based on the MCP protocol, with a built-in real-time web monitoring panel.
Use natural language to have the AI assistant communicate directly with your microcontroller and embedded devices.

✨ Functional Features

🖥️ MCP Support — Seamless integration with AI clients via Model Context Protocol
🔌 Serial Port Control — Automatically scans, connects, and reads serial port devices
🌐 Web Monitoring Panel — Built-in HTTP + WebSocket dual-port service, real-time browser bypass monitoring
Compatible with Windows, MacOS, Linux

🛠️ Available tools

Tools	Instructions
`list_ports`	Scan all available serial ports on the machine
`connect_port`	Connect to a specified serial port (supports custom baud rate, default 115200)
`close_port`	Explicitly disconnect the current serial port connection
`write_data`	Write data to the serial port (auto-complete line new)
`read_data`	Read serial port buffer data (including user intervention history)
`start_monitor_ui`	Launch the Web Monitoring Panel (default HTTP 8080 / WebSocket 8081)

📦 Directly Configure into AI Assistants (Recommended)

This project has been packaged and uploaded to the uv and pypi warehouses. You can add it directly from the following command line in major AI assistants, and it will automatically pull and configure it.

Add via command line

Using mcp.json Add (Kimi Code)

Kimi Code configures the MCP server via mcp.json. Edit the user-level configuration ~/.kimi-code/mcp.json (or item-level .kimi-code/mcp.json), and add the following:

You can also run /mcp-config in Kimi Code’s TUI to interactively add it; After configuration, run /mcp to view connection status.

Adding with toml (Codex)

This is the default and recommended configuration format for Codex CLI. Open your configuration file (such as ~/.codex/config.toml) and add the following:

Add via JSON

🦞OpenClaw, Cursor, Cline, TRAE, Cherry Studio, Qwen Chat, and other clients, in the MCP server settings, choose to use JSON to add and paste the following fields.

📦 Manually download, install, and run

If you want to customize and modify this tool,

Method 1: Install via Python

Or use pip:

Method 2: Local Development and Installation

1. Clone warehouse

2. Installation dependency

Ensure UV is installed:

3. Local operation

This page was translated automatically. Technical details may differ from the original.

Install

uvx --from umeko-serial-mcp start-serial-mcp

New to MCP? Start with the mcpskill.org guide →

Configuration

{
  "mcpServers": {
    "serial-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "umeko-serial-mcp",
        "start-serial-mcp"
      ]
    }
  }
}