Gpu Check

Automatically polls and displays memory usage and online status of RTX 3090 and 4090 AI compute nodes in the local network.

Security Scan

The skill's code and instructions match its stated purpose (polling two local GPU nodes) and request no extra credentials or external endpoints, so its footprint is internally consistent.

Purpose & Capability

Name/description say it polls local RTX 3090/4090 nodes; the code (JS and Python) specifically queries the two private IPs on port 5000 and renders usage—this aligns with the stated purpose. Dependency on axios (for Node) is appropriate.

Instruction Scope

SKILL.md instructs installing axios and describes Node usage which matches gpu_check.js. Minor inconsistencies: a Python implementation (gpu_check.py) is included but SKILL.md does not mention Python or the requests library; package.json lists main:index.js which does not exist. These are quality issues but do not indicate malicious behavior.

Install Mechanism

No automated install/spec is provided (instruction-only). The package.json and package-lock.json reference npm packages from the public registry (axios and its deps), which is standard and expected.

Credentials

The skill requests no environment variables, no credentials, and only accesses two local private IP endpoints that match the stated purpose. Network access to those IPs is necessary for functionality.

Persistence & Privilege

Skill does not request always:true or other elevated platform privileges and does not modify other skills or system-wide config. Autonomous invocation is allowed by default (normal).

Assessment

This skill appears to do only what it claims: poll two local IPs for GPU memory usage. Before installing, verify the two IP addresses (192.168.2.236 and 192.168.2.164) are your intended GPU nodes. Note small inconsistencies (a Python implementation is included but not documented; package.json points to index.js which is missing). Run npm install only in a controlled environment (or inspect gpu_check.js/gpu_check.py yourself). If you don't use Python, you can ignore or remove gpu_check.py; if you don't want Node dependencies, avoid running npm install. If these nodes are unknown or unexpected on your network, do not install or run the skill.

GPU 状态检查 (gpu_check)

实时获取局域网内分布式 AI 算力节点的显存占用情况。

功能

• 自动轮询 3090 (192.168.2.236) 和 4090 (192.168.2.164) 的显存状态
• 输出带进度条的 Markdown 表格
• 监控各节点 API 服务在线情况

依赖

• Node.js 环境（已内置）
• axios 库（需安装）

安装

1. 在技能目录安装依赖：

   cd ~/.openclaw/workspace/skills/gpu_check
   npm init -y
   npm install axios
   

2. 确保 GPU 节点 API 已启动（需在 192.168.2.236 和 192.168.2.164 运行支持 /gpu 端点的服务）

使用

在聊天中发送：

• /gpu
• @机器人 显卡状态
• 查看 GPU 占用