← 返回 Skills 市场
5843
总下载
4
收藏
34
当前安装
2
版本数
在 OpenClaw 中安装
/install screen-monitor
功能描述
Dual-mode screen sharing and analysis. Model-agnostic (Gemini/Claude/Qwen3-VL).
安全使用建议
Review before installing. Use this only on trusted networks, share a single non-sensitive window when possible, stop sharing when finished, and assume screenshots may be written locally under /tmp. Avoid using it where passwords, customer data, private messages, admin consoles, or regulated information may appear on screen.
功能分析
Type: OpenClaw Skill
Name: screen-monitor
Version: 1.0.1
The skill is designed for screen sharing and analysis by an AI agent. It uses a local Node.js server (`references/backend-endpoint.js`) on port 18795 to receive screen frames via WebRTC from `web/screen-share.html`, saving them to `/tmp/clawdbot-screen-latest.png`. The `references/screen-analyze.sh` script then instructs the local `clawdbot agent` to analyze this image. While the WebRTC portal is accessible via LAN IP (`references/get-share-url.sh`) and declared as 'remote_access' in `manifest.json`, the user must explicitly initiate screen sharing, and all captured data is processed locally by the agent, not exfiltrated to external unauthorized endpoints. There is no evidence of malicious intent, unauthorized execution, or prompt injection attempts to subvert the agent's core directives.
能力评估
Purpose & Capability
Screen sharing, screenshot capture, frame storage, and vision analysis are coherent with a screen-monitoring skill, and no artifact shows exfiltration, credential theft, destructive behavior, or deception.
Instruction Scope
The instructions disclose screen sharing and analysis, but they do not clearly warn that frames are uploaded every 3 seconds, written to /tmp, and that the analysis script may take a full desktop screenshot when no portal frame exists.
Install Mechanism
Installation is ordinary for a skill package and declares node, curl, a fixed backend port, remote_access, and optional ImageMagick for the screenshot fallback. No suspicious installer or hidden dependency behavior was found.
Credentials
The backend listens on a fixed port, the share URL uses the LAN IP, CORS allows any origin, and the API has no authentication. That is broader than necessary for a privacy-sensitive screen capture workflow.
Persistence & Privilege
Captured frames are persisted to predictable /tmp paths, and the fallback can capture the full OS desktop. These are high-sensitivity operations without clear retention controls or per-capture consent.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install screen-monitor - 安装完成后，直接呼叫该 Skill 的名称或使用
/screen-monitor触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
- Updated manifest.json and README.md (SKILL.md) for minor improvements.
- No functional changes to the skill or usage instructions.
v1.0.0
- Initial release of screen-monitor with dual-mode screen sharing and analysis.
- Supports Fast Share via local WebRTC portal for quick visual checks.
- Enables Full Control mode using a browser extension for deep debugging and UI automation.
- Compatible with Gemini, Claude, and Qwen3-VL vision models.
- Provides easy setup steps and describes key files and ports used.
元数据
常见问题
Screen Monitor 是什么？
Dual-mode screen sharing and analysis. Model-agnostic (Gemini/Claude/Qwen3-VL). 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 5843 次。
如何安装 Screen Monitor？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install screen-monitor」即可一键安装，无需额外配置。
Screen Monitor 是免费的吗？
是的，Screen Monitor 完全免费（开源免费），可自由下载、安装和使用。
Screen Monitor 支持哪些平台？
Screen Monitor 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Screen Monitor？
由 emasoudy（@emasoudy）开发并维护，当前版本 v1.0.1。
推荐 Skills