screenshot-mcp
Enables AI agents to capture screenshots of windows (e.g., WeChat developer tools) and save them locally, allowing the agent to read the image directly and close the code-change-to-review loop without OSS or network.
README
mcp-screenshot
跨平台 MCP server,让 AI agent 自己截取微信开发者工具模拟器画面,形成「改代码 → 截图 → 看效果」闭环。
它解决什么问题
以前 agent 改完小程序看不到效果,要人工截图 → 传 OSS → 给链接。本 MCP 把这一环自动化:agent 调一个工具截当前画面,拿到本地路径后自己 Read 读取,不依赖 OSS、不依赖网络。
核心思路:agent 本身能读本地图片,所以 MCP 的职责只是「截图 → 存本地 → 返回绝对路径」,剩下的闭环 agent 自己完成。
工具
screenshot_capture
| 参数 | 类型 | 说明 |
|---|---|---|
| scope | "window" | "region" | "full" |
默认 window。window 按标题找;region 按坐标;full 全屏兜底 |
| title | string | 窗口标题关键词,默认匹配微信开发者工具相关窗口 |
| region | string "x,y,w,h" |
仅 scope=region 时生效,四个整数像素坐标 |
成功返回本地 PNG 绝对路径,agent 用 Read 工具读取该路径即可看到画面。失败返回结构化错误 + 可读的 hint(例如建议改用 scope=full 兜底)。
接入 ZCode
在项目工作区 .zcode/config.json(配置文件不展开 ${...} 模板变量,必须用绝对路径):
{
"mcp": {
"servers": {
"screenshot": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/mcp-screenshot/dist/index.js"]
}
}
}
}
重启会话,输入 /mcp 查看连接状态(应显示 connected)。截图保存在项目 .screenshots/(已自动加入 gitignore)。
备注:
projectRoot默认用 server 进程的process.cwd();如果 ZCode 启动 server 时的 cwd 不是你期望的项目根,可设置环境变量MCP_SCREENSHOT_PROJECT_ROOT指向目标项目目录,截图就会存到那里的.screenshots/。
开发
cd mcp-screenshot
bun install
bun run build # 编译到 dist/(tsc,零错误)
bun test # 跑测试(bun test)
bun run start # 等价于 node dist/index.js
源码结构:
src/
├── index.ts # stdio 入口
├── server.ts # MCP 工具定义(screenshot_capture)
├── paths.ts # 截图路径/gitignore 维护(纯函数,已测)
└── platform/
├── types.ts # 平台捕获接口
├── index.ts # 按 process.platform 分发
├── win32.ts # Windows: PowerShell + System.Drawing
└── darwin.ts # macOS: screencapture + osascript
平台支持
| 平台 | 实现 | 状态 |
|---|---|---|
| Windows | PowerShell + System.Drawing.CopyFromScreen,按窗口标题定位 | 已实测 |
| macOS | screencapture -l<wid> / -R x,y,w,h,osascript 查窗口 |
代码就绪,待 Mac 实测(需授予「屏幕录制」权限) |
设计原则
- capture() 永远 resolve,绝不 reject——失败也通过结构化的
CaptureFailure表达。stdio MCP server 一旦 crash,整个会话的工具都没了,所以稳定性是第一约束。 - 窗口找不到是正常情况——返回清晰错误 + hint,让 agent 自己判断(例如改用
scope=full兜底)。 - YAGNI——MVP 只截当前画面,不做点击/导航/OCR/对比。
后续规划
- v2:
screenshot.list_windows列出所有窗口,让 agent 精确指定截哪个 - v3: 接 miniprogram-automator,加 navigate/tap,真正"控制+截图"闭环
- v4: CDP 直连模拟器 Chromium,截纯渲染画面
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.