Allows using a local vision model (e.g., minicpm-v) via Ollama as a backup or alternative to the cloud API.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Vision MCP Server describe my current screen"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
截图 + 视觉模型分析 MCP 服务器，支持 Claude Code 等 MCP 客户端。
工具 | 说明 |
| 截图并用视觉模型分析（支持全屏/主屏/指定窗口） |
| 纯截图保存，不分析 |
| 列出当前所有可见窗口标题 |
| 分析一张已有的图片文件 |
Related MCP server: Screen View MCP
claude /plugin install github.com/你的用户名/vision-mcp-server在 ~/.mcp.json 中添加：
{
"mcpServers": {
"vision": {
"command": "node",
"args": ["路径/server.mjs"],
"env": {
"DASHSCOPE_API_KEY": "你的阿里云百炼 API Key"
}
}
}
}Node.js ≥ 18
npm ≥ 9
阿里云百炼 API Key：去 bailian.console.aliyun.com → API Key 创建
Windows：支持（PowerShell + .NET）
可选：Ollama + minicpm-v:8b（本地模型备用）
环境变量 | 必填 | 默认值 | 说明 |
| ✅ | - | 阿里云百炼 API Key |
| ❌ |
| 云端模型名 |
| ❌ |
| 本地备用模型 |
| ❌ |
| 截图保存目录 |
"帮我看看浏览器当前页面"
"截一张全屏截图"
"分析这张图片：C:\photo.jpg"
"现在有哪些窗口开着"This server cannot be deployed
Capture screenshots, detect visual regressions between page versions, and analyze with AI.
Screenshot any URL/HTML as PNG/JPEG/WebP, or read it as clean Markdown/text for LLMs.
Generate images, GIFs, and PDFs from HTML, URLs, or templates — from your AI agent.
Screenshot, PDF and HTML-to-image rendering API so Claude and Cursor can see any web page.
