ClawHub

Xiaozhi Mcp Openclaw Official

MCP Tools

按小智官方 MCP 接入方式，把小智 AI 设备通过 MCP 接到 OpenClaw / OpenAI-compatible 后端。适用于已经有小智 MCP 接入点（wss://api.xiaozhi.me/mcp/?token=...）的场景。提供一个 `openclaw_query(message)` MCP 工具，让小智在需要外部能力、复杂推理、联网查询或外部智能辅助时调用。 Official XiaoZhi MCP bridge for OpenClaw / OpenAI-compatible backends. Use when you already have a XiaoZhi MCP endpoint and want XiaoZhi to call an external assistant tool such as `openclaw_query(message)`.

Install

openclaw skills install xiaozhi-mcp-openclaw-official

xiaozhi-mcp-openclaw-official

简体中文 | English

简体中文

作用

这是一个按小智官方文档方式接入的最小可用 MCP 原型。

数据流：

小智
→ MCP 接入点
→ mcp_pipe.py
→ openclaw_mcp.py
→ OpenAI/OpenAI-compatible 后端
→ 回复文本
→ 小智播报

提供的工具

  • openclaw_query(message)
    • 当用户提到“龙虾机器人”“龙虾”或需要外部能力时，调用这个工具
    • 返回适合语音播报的简短中文结果

启动步骤

  1. 安装依赖：
pip install -r requirements.txt
  1. 复制配置：
cp .env.example .env
  1. 修改 .env：
  • MCP_ENDPOINT
  • OPENAI_BASE
  • OPENAI_KEY
  • MODEL
  1. 启动：
python3 mcp_pipe.py openclaw_mcp.py

关键配置项

  • MCP_ENDPOINT：小智 MCP 接入点
  • OPENAI_BASE：OpenAI/OpenAI-compatible 后端地址
  • OPENAI_KEY：后端 key
  • MODEL：模型名（如 gpt-5.4）

注意事项

  • 不要把真实 .env 连同密钥一起分享
  • 这是一个原型桥接项目，后续可以继续增强成真正的 OpenClaw 多工具接入版

English

Purpose

This is a minimal working MCP prototype that follows XiaoZhi's official MCP integration style.

Flow:

XiaoZhi
→ MCP endpoint
→ mcp_pipe.py
→ openclaw_mcp.py
→ OpenAI/OpenAI-compatible backend
→ reply text
→ XiaoZhi speaks it

Provided tool

  • openclaw_query(message)
    • Use when the user mentions “龙虾机器人 / Lobster Robot” or when an external capability is needed
    • Returns a short Chinese reply suitable for voice playback

Startup

  1. Install dependencies:
pip install -r requirements.txt
  1. Copy config:
cp .env.example .env
  1. Update .env:
  • MCP_ENDPOINT
  • OPENAI_BASE
  • OPENAI_KEY
  • MODEL
  1. Start:
python3 mcp_pipe.py openclaw_mcp.py

Key config values

  • MCP_ENDPOINT: XiaoZhi MCP endpoint
  • OPENAI_BASE: OpenAI/OpenAI-compatible backend URL
  • OPENAI_KEY: backend API key
  • MODEL: model name (for example gpt-5.4)

Notes

  • Do not publish your real .env or secrets
  • This is a prototype bridge and can later be extended into a richer OpenClaw multi-tool integration
More in MCP Tools