draw-guess-mcp
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., "@draw-guess-mcpmy guess is a star"
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.
draw-guess-mcp
让非多模态 AI 也能玩「你画我猜」的 MCP server 骨架。
核心不是让模型真的看见图片,而是把画变成结构化线条(SVG 路径 + ASCII 网格), 通过 MCP 工具让模型 读 / 画 / 猜。
思路整理自教程《让非多模态 AI 也能玩你画我猜》。
两种玩法
玩法 A:AI 画,用户猜。AI 调 draw_start 传入答案与线条数组,前端展示 SVG,用户猜。
玩法 B:用户画,AI 猜。用户在前端画图,后端压成 SVG + ASCII 给 AI,AI 调 draw_status 读取后 draw_guess 提交。(本骨架当前完整实现玩法 A;玩法 B 见文末扩展。)
Related MCP server: Excalidraw MCP Server
三个 MCP 工具
工具 | 作用 | 是否含答案 |
draw_start(answer, content, aliases, artist) | AI 开始一题,生成 SVG + ASCII | 输入含答案,但不回传 |
draw_status(round_id?) | 查看当前画作(SVG + ASCII 网格) | 绝不返回答案 |
draw_guess(guess, guesser?, round_id?) | 提交猜测 | 只回猜对了 / 没猜中,不剧透 |
运行
pip install -r requirements.txt
python draw_guess_server.py
# 客户端连 http://localhost:3456/mcp本地 stdio 接入(Claude Desktop 等):把 main 里换成 mcp.run()。
Docker
不想自己管 Python 环境,用 Docker 一键起常驻服务:
docker compose up -d
# 客户端连 http://localhost:3456/mcp服务容器内监听 0.0.0.0:3456(本仓库已设置),映射到宿主 3456。 停止:docker compose down
示例(玩法 A)
调用 draw_start,传入一颗五角星的 polyline(具体 payload 见 examples/build_star.py,运行它会打印可直接粘贴的参数)。 draw_status 会返回 drawing_svg 与 ascii_grid,但没有 answer。
扩展点
多人历史:已用 guesses 列表记录每人的猜测
别名答案:draw_start 的 aliases 已支持
难度模式:draw_start 加 max_lines,截断 strokes[:N]
回放模式:把 content 按笔画顺序存下来
玩法 B:加 submit_user_drawing(svg/ascii) 工具,AI 调 draw_status 来猜
部署 / Deploy
server 监听端口可用环境变量 PORT 覆盖(默认 3456),host 固定 0.0.0.0,方便上云。
Hugging Face Spaces(Docker SDK,免费)
新建 Space,SDK 选 Docker。
上传本仓库文件(含 Dockerfile)或关联此 GitHub 仓库。
HF 注入 PORT=7860,server 自动监听;Space 公开后地址形如 https://-.hf.space/mcp 。
Render(免费)
仓库根已有 Dockerfile。新建 Web Service 连本仓库,Runtime 选 Docker,部署后 Render 注入 PORT,地址形如 https://draw-guess-mcp.onrender.com/mcp 。
Railway
仓库根有 railway.json。导入 GitHub 仓库即部署,Railway 注入 PORT。
Fly.io
仓库根有 fly.toml。执行 fly launch / fly deploy 即可,已配置监听 PORT=8080。
License
MIT
This server cannot be deployed
Maintenance
Related MCP Connectors
Create and edit architecture diagrams from your AI agent; get an SVG and a live editable canvas.
- bluffnetOAuthgg.bluffnet
Live Texas Hold'em for AI agents. The tools teach the rules; the bluffing is up to your model.
Generate vector art, vectorize images, and return SVG, PNG, and logo kits to AI agents.
Games for AI agents. Each game runs inside a single context window.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables creation, management, and export of Excalidraw drawings through natural language. Supports CRUD operations on drawings and export to SVG, PNG, and JSON formats with file-based storage.82,102 npm-
- AlicenseAqualityCmaintenanceEnables AI agents to programmatically generate, edit, and view Excalidraw diagrams with real-time browser synchronization. It provides a suite of tools for adding shapes, text, and arrows to diagrams through natural language interactions.112,102 npm100MIT
- FlicenseAqualityCmaintenanceExposes ASCIIFlow drawing primitives as tools to enable AI assistants to generate ASCII wireframes and diagrams from natural language descriptions. It provides commands for creating canvases and drawing boxes, lines, arrows, and text using a character-grid coordinate system.81-
- AlicenseNot gradedqualityDmaintenanceEnables Claude to draw on a live tldraw canvas from prompts, with real-time visualization in a browser. Provides tools for creating, updating, deleting shapes and retrieving canvas state.14MIT