MLX Local AI OpenClaw Skill
一键部署支持Apple Silicon的本地LLM与中文向量Embedding服务，含统一API网关及便捷启动管理脚本。
Installation
clawhub install mlx-local-ai
Requires npm i -g clawhub
234
Downloads
0
Stars
1
current installs
1 all-time
1
Versions
EU-Hosted Inference API Power your OpenClaw skills with
the best open-source models.
Drop-in OpenAI-compatible API. No data leaves Europe.
Explore Inference APIGLM
GLM 5
$1.00 / $3.20
per M tokens
Kimi
Kimi K2.5
$0.60 / $2.80
per M tokens
MiniMax
MiniMax M2.5
$0.30 / $1.20
per M tokens
Qwen
Qwen3.5 122B
$0.40 / $3.00
per M tokens
MLX Local AI 技能
一键部署本地 AI 服务，支持 Apple Silicon (M1/M2/M3/M4)。
功能
- 🤖 本地 LLM 服务：Qwen3.5-4B-OptiQ-4bit (通过 MLX-LM)
- 📊 Embedding 服务：bge-base-zh-v1.5 (中文向量模型)
- 🔄 OpenClaw Gateway：统一 API 网关
- 🎯 一键安装/启动/停止
系统要求
- macOS 14.0+ (Sonoma 或更高)
- Apple Silicon (M1/M2/M3/M4)
- Python 3.10+
- 8GB+ 内存
快速开始
安装
./install.sh启动服务
./start_ai.sh start检查状态
./start_ai.sh status停止服务
./start_ai.sh stop服务端口
| 服务 | 端口 | 说明 |
|---|---|---|
| Chat API | 8080 | OpenAI 兼容 API |
| Embedding API | 8081 | 向量嵌入 API |
| OpenClaw Gateway | 18789 | 管理界面 |
API 使用
Chat API
curl http://localhost:8080/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "mlx-community/Qwen3.5-4B-OptiQ-4bit",
"messages": [{"role": "user", "content": "你好"}],
"max_tokens": 100
}'Embedding API
curl http://localhost:8081/embed \
-H "Content-Type: application/json" \
-d '{"texts": ["你好", "世界"]}'文件结构
mlx-local-ai/
├── SKILL.md # 本文件
├── install.sh # 安装脚本
├── start_ai.sh # 启动脚本
├── uninstall.sh # 卸载脚本
├── config/ # 配置文件
│ ├── openclaw.json # OpenClaw 配置
│ └── env.example # 环境变量示例
└── scripts/ # 工具脚本
├── check_deps.py # 依赖检查
└── test_api.py # API 测试配置
环境变量
复制 config/env.example 到 ~/.zshrc：
cp config/env.example ~/.zshrc.local
echo 'source ~/.zshrc.local' >> ~/.zshrc模型配置
默认使用 mlx-community/Qwen3.5-4B-OptiQ-4bit，可在 start_ai.sh 中修改。
故障排查
服务无法启动
# 检查依赖
python3 scripts/check_deps.py
# 查看日志
./start_ai.sh logs模型加载失败
# 手动下载模型
source ~/mlx-env/bin/activate
HF_ENDPOINT=https://hf-mirror.com python3 -c "from mlx_lm import load; load('mlx-community/Qwen3.5-4B-OptiQ-4bit')"更新
./install.sh --update卸载
./uninstall.sh许可证
MIT License
Statistics
Author
saybanet
@saybanet
Latest Changes
v1.0.1 · Mar 9, 2026
Initial release of mlx-local-ai: one-click local AI services for Apple Silicon. - Provides local LLM (Qwen3.5-4B-OptiQ-4bit via MLX-LM) and embedding (bge-base-zh-v1.5) services - Unified OpenClaw API gateway with OpenAI-compatible endpoints - Simple installation, start/stop scripts, and status checking - Clear API usage examples and troubleshooting steps - macOS 14+, Apple Silicon, Python 3.10+ required
Quick Install
clawhub install mlx-local-ai Related Skills
Other popular skills you might find useful.
Made in Europe Chat with 100+ AI Models in one App.
Use Claude, ChatGPT, Gemini alongside with EU-Hosted Models like Deepseek, GLM-5, Kimi K2.5 and many more.
