Yahboom Mcp

io.github.sandraschi由 GitHub 账号 sandraschi 发布v2.5.0b1更新于 Oct 27, 2026

SOTA 2026 Yahboom Raspbot v2 ROS 2 MCP Server

概览

AI 生成的概览

让 AI 助手通过 ROS 2 控制 Yahboom Raspbot v2(Boomy)机器人小车的 MCP 服务器,可执行任务、读取遥测与诊断。

可以做什么
  • 控制机器人:通过 MCP 工具和通用工具操作驱动麦轮底盘、灯带、OLED 等外设。
  • 执行智能体任务:把自然语言目标转成计划并在机器人上执行,含视觉检测与超声波避障。
  • 检查运行栈:浏览 ROS 话题与节点、查看栈健康状态、读取日志并进行基于 SSH 的诊断。
  • 配合网页仪表盘进行人工遥测观察与外设控制,与机器接口并存。
适用场景
适合已拥有 Yahboom Raspbot v2、希望让 AI 智能体驾驶机器人、执行规划任务或从桌面 MCP 客户端检查其 ROS 2 运行栈与健康状态的用户。
运行要求
以 stdio 方式本地运行;README 给出通过 uvx 从 Git 仓库安装、Claude Desktop 的 MCPB 包以及 Windows 安装程序。需要能访问机器人网络,通过 YAHBOOM_IP 环境变量配置,并具备可用的 ROS 2 / rosbridge 或 ESP32 连接。可选的 LLM 规划使用 Ollama、LM Studio 或 Gemini。
安装前请注意
Windows 安装程序未签名,SmartScreen 会给出警告。服务器可驱动实体机器人并执行自主任务,需确保周围环境安全并有人监督。它可建立 SSH 会话并读取容器日志,还会通过网络访问机器人及已配置的 LLM 服务商。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Yahboom Mcp,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

Yahboom Raspbot v2 (Boomy) — MCP Server, Webapp & Documentation

[Just] [Ruff] [Python] [Biome] [FastMCP]

📖 Installation Guide — quick start, manual setup, and troubleshooting

[CI] [FastMCP 3.4] [Python 3.12+] [ROS 2 Humble] [Ruff] [Biome] [License] [Stars] [Last commit] [Ollama] [LM Studio] [Topics]

[Boomy Hero Shot]

Download for Windows: yahboom-mcp-2.5.0-setup.exe — installer with webapp + embedded backend, one shortcut. The installer is unsigned: SmartScreen will warn — click More info > Run anyway. Claude Desktop users can instead install the MCP bundle (one-liner below).

Quick Start

Claude Desktop only (MCPB bundle):

powershell
irm https://github.com/sandraschi/yahboom-mcp/releases/latest/download/install.ps1 | iex

Every other client (Cursor, Windsurf, OpenCode, Claude Code) — stdio via uvx:

json
{  "mcpServers": {    "yahboom-mcp": {      "command": "uvx",      "args": ["--from", "git+https://github.com/sandraschi/yahboom-mcp", "yahboom-mcp", "--mode", "stdio"],      "env": { "YAHBOOM_IP": "192.168.1.11" }    }  }}

Developers from source:

powershell
git clone https://github.com/sandraschi/yahboom-mcpcd yahboom-mcpjust

This opens an interactive dashboard showing all available commands. Run just bootstrap to install dependencies, then just serve or just dev to start.

Manual Setup

If you don't have just installed:

Overview

Boomy is a Yahboom Raspbot v2 (Mecanum Version) robot car. Weighing in at approximately 1.0 kg and costing roughly $300 (fully loaded with a Raspberry Pi 5 16GB), it provides a unified platform for testing sensor integration, ROS 2 orchestration, and agentic (AI-driven) robotics. It's also great fun for kids of all ages.

The Manufacturer: Yahboom

Yahboom is a prominent player in the Chinese robotics and STEM education industry. They offer a diverse range of hardware platforms:

  • Market Range: From entry-level $50 DIY kits to advanced $2,000+ STEM robots with precision mechanical arms.
  • Industry Position: They specialize in accessible hardware for education and prototyping, positioned below enterprise-grade platforms such as Unitree or Noetix.
  • Developer Ecosystem: Yahboom maintains a strong commitment to a FOSS (Free and Open Source Software) stack and provides active GitHub support for developers.

🏗️ Hardware Architecture

Boomy utilizes a "Double-Stack" controller model to bridge real-time physical reliability with high-level cognitive tasks:

  1. Rosmaster (ESP32 / Micro-ROS): Small-form-factor hardware controller for low-latency motor control, ultrasonic pinging, and line-sensor processing.
  2. Raspberry Pi 5 (Optional / Gateway): The main brain for standalone operation.

🚀 The "Zero-Host" Swarm Config

For scaled deployments, Boomy can operate in a Pi-less Swarm mode:

  • Setup: Remove the Raspberry Pi; replace with a WiFi-to-Ethernet/UART bridge ($15).
  • Control: Centralized orchestration via a remote Mothership (PC with RTX 4090).
  • Scaling: Deploy 3+ units for the price of a single Pi-hosted robot.

🕹️ Interface Separation

This project clearly distinguishes between human-operable controls and machine-optimized interfaces:

  • Human Interface: A browser-based Web Dashboard (React/Vite) designed for manual telemetry observation, peripheral control (lights, OLED), and human-robot interaction.
  • Machine Interface: A standards-compliant MCP Server (Model Context Protocol) designed for AI agents and MCP clients to programmatic control Boomy as a tool within a larger fleet.

Fleet Crossconnects (Companions)

yahboom-mcp works completely standalone. You can optionally connect it with companion servers in the sandraschi fleet to unlock extended features:

Companion ServerFeature UnlockedStatusSetup
speech-mcpspoken-feedback, voice-synthesisOptionalInstall Guide
reaper-mcpaudio-playback, soundboard-fxOptionalInstall Guide
virtualdj-mcpdeck-control, music-playbackOptionalInstall Guide

Self-Contained Companions: Fleet companions operate independently. Installing companions does not trigger transitive dependency chains.


📂 Documentation Pillars

PillarFocusKey Topics
Autonomous Missionsv2.4.0 — Ollama → ROSNatural-language goals → LLM planning → ROS execution; vision detection (SSD MobileNet v2); ultrasonic obstacle avoidance; "find our dog" walkthrough.
Startup & bringupStart here (robot + Goliath)ROS 2 vs rosbridge_suite (software) vs USB controller tier under the Pi; boot order; Docker/systemd on the Pi; what the webapp can restart; dashboard vs diagnostics.
Agent missions & MCPLLM → Pi missionsyahboom_agent_mission (MCP) and POST /api/v1/agent/mission (HTTP); MissionPlanV1; env vars; boomy_mission_executor; Nav2 and /boomy/detections_json; troubleshooting vs yahboom_agentic_workflow.
Rosmaster ArchitectureDual-bus analysisI2C (0x2b) for motors/sensors, UART for IMU/battery; register map; container layout; micro-ROS analysis.
Stack health probehealth.stackTTL SSH/TCP probes, YAHBOOM_ROS2_CONTAINER, lifecycle (never started vs exited vs restart loop), optional docker logs preview (redacted).
Setup & InstallationDev installuv, modes, baseline env vars.
Software & LogicArchitectureSystem design, ROS 2 node graphs, and state management.
Hardware & PinoutsPhysical LayerWiring diagrams, I2C addresses, and sensor technical specs.
Raspbot v2 hardware stackBoomy chassisChassis, mecanum, motors, lightstrip, PTZ, camera, IMU, ultrasonic, line/cliff, expansion board vs rosbridge software, Pi ports, I2C, battery, switch.
ROS 2 BridgeConnectivityBridge architecture, topic map, state cache, env vars, known bugs.
Voice & AudioSound SystemCSK4002 module protocol, espeak-ng TTS, chatrobot architecture, audio soundboard with 17 built-in effects.
Multi-Robot IntegrationEcosystemFederated fleet standards and cross-robot communication protocols.
ConfigurationReferenceEnv vars, ports, transport modes, LLM providers.
DevelopmentReferenceDev setup, layout, test commands, five-gate.
ToolsReferenceMCP tools, yahboom_tool operations, REST endpoints.
TroubleshootingReferenceCommon issues and fixes.
OnboardingFirst-time setupHardware bring-up for Boomy (robot power, IP, sanity checks).

Webapp status surfaces

  • Dashboard (/dashboard) — First page: basic robot link (gateway, target IP, ROS · SSH · video) and operator hints when the Raspbot is off or Goliath has no AP/Ethernet path.
  • Diagnostic Hub (/diagnostics) — Detail: ROS topic explorer, node list, resync / hard reset, SSH shell, stack health table (same stack payload as the dashboard when available).
  • Server logs (/logs) — Live yahboom-mcp log stream (SSE) on Goliath.

Historical status logs and legacy research are preserved in the docs/archive/ directory.

Stack

LayerTech
MCPFastMCP 3.4 (Unified Gateway: stdio + HTTP/SSE)
BackendFastAPI + uvicorn, Pydantic v2, structlog
Robot linkroslibpy (rosbridge) / direct ESP32 TCP / mock bridge, paramiko SSH
Vision/audioOpenCV, faster-whisper, edge-tts, pytapo
FrontendReact 19 + Vite 5 + TailwindCSS 3 + Lucide + Framer Motion + Zustand + Three.js
Package managersuv (Python), bun (webapp)
DesktopTauri 2 + NSIS (PyInstaller sidecar backend)
Qualityruff + pyright + pytest (+coverage) / tsc + Biome / Playwright
LLMOllama (:11434) + LM Studio (:1234) auto-discovery, Gemini mission planning

🛡️ Industrial Quality Stack

This project adheres to SOTA 2026 industrial standards for high-fidelity agentic orchestration:

  • Python (Core): Ruff for linting and formatting. Zero-tolerance for print statements in core handlers (T201).
  • Webapp (UI): Biome for sub-millisecond linting.
  • MCP: FastMCP 3.4 with Unified Gateway, portmanteau tools, skills, and agentic sampling.
  • LLM Providers: Auto-discovery of Ollama (:11434) and LM Studio (:1234). GPU metrics via nvidia-smi.
  • Automation: Justfile recipes for all fleet operations (just lint, just fix, just dev).

来源:README.md,提交 58378e3

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

1
  1. v2.5.0b1最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分