sketch/design — System Design Interview Board

io.github.tGhattas由 GitHub 账号 tGhattas 发布v0.1.2更新于 Oct 30, 2026

Hand-drawn whiteboard for system design mock interviews; your AI watches, asks and gives feedback.

概览

AI 生成的概览

本地手绘白板,让 AI 充当系统设计面试官:观察你的架构图、语音提问并给出评分反馈。

可以做什么
  • 开展面试会话:设定题目、计时与阶段,循环等待候选人说话或白板变化,再用简短口语化消息回应。
  • 读取白板:获取结构化的组件、连线、区域、便签及变更,或整块白板的 PNG 图片。
  • 标注设计:在组件上钉面试官评论气泡、高亮并缩放定位,也可清除这些标注。
  • 在白板上作画:以同样的手绘风格添加组件、连线、便签和区域。
  • 推进与收尾:切换阶段追踪器、导出完整聊天与语音记录,并以 1-5 分反馈卡结束面试。
适用场景
适合练习系统设计模拟面试:你在白板上画设计,让 AI 追问需求、规模估算、API、数据模型、瓶颈与取舍,最后给出评分反馈。不接 AI 时也可当作普通手绘白板使用。
运行要求
需要 Node.js 18+,通过 npx 启动本地进程(npm 包 sysdesign-board)。需要一个支持 MCP 的 AI 应用,如 Claude Desktop、Claude Code、Codex CLI、ChatGPT 或其他客户端。白板在浏览器中打开,默认端口 4321,可用 SYSDESIGN_PORT 修改。语音输入需 Chrome、Edge 或 Safari 并授权麦克风。无需 API 密钥或账号。
安装前请注意
白板描述、白板图片、聊天消息和语音记录会发送给你所连接的 AI 应用,受该服务商隐私政策约束。浏览器语音识别会把音频发送给 Google(Chrome)或 Apple(Safari)。可选的 ChatGPT 隧道会把 /mcp 端点公开到互联网,其带令牌的网址应视同密码。面试官可在你的白板上作画和标注;会话数据保存在本地,直到你清空白板或删除数据目录。

安装

在 SourceWeft 中

  1. 打开 控制台中的 sketch/design — System Design Interview Board,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

sketch/design — the system design interview board

A hand-drawn whiteboard for system design mock interviews, plus an MCP server that turns Claude or ChatGPT into your interviewer. It sees your board, listens to you, talks back, pins questions on your components, and ends with scored feedback.

[A mock interview in progress: a URL shortener diagram, an interviewer question pinned on Redis, and the voice/text chat]

  • Pencil-style canvas: a paper or chalkboard theme, with shapes that look drawn by hand (rough.js + perfect-freehand)
  • 27 system design components: load balancer, API gateway, CDN, cache, SQL and NoSQL databases, object store, queue, Kafka-style stream, pub/sub, workers, serverless functions, auth, monitoring, ML and more
  • Arrows that stay attached to components, with labels, curved or elbow routes, and dashed lines for async flows. Hover a component and drag its blue dot to connect it.
  • Interview panel: the problem, a timer, a phase tracker, chat, voice in (speech-to-text) and voice out (the interviewer's replies are read aloud)
  • The interviewer can read your diagram as structured data, look at a picture of it, pin comments on components, highlight things, sketch reference designs, and grade you at the end
  • Runs entirely on your machine. The AI runs in your own Claude or ChatGPT app, on your existing subscription, with no API keys.

Install

Requires Node.js 18+.

Your AI appHow to connectBilled to
Claude DesktopDouble-click sysdesign-board-x.y.z.mcpb from Releases, or run npx -y sysdesign-board installClaude Pro / Max
Claude Codeclaude mcp add sysdesign-board -- npx -y sysdesign-boardClaude Pro / Max
Codex CLIAdd it to ~/.codex/config.toml (below)ChatGPT Plus / Pro
ChatGPT (desktop / web)Developer Mode connector through a tunnel (below)ChatGPT Plus / Pro
Cursor, Windsurf, VS Code, othersThe standard MCP JSON (below)Your plan there

The .mcpb one-click file is Claude Desktop only. Every other app runs the same npm package.

Claude Desktop

Option A: one-click. Download the .mcpb file from Releases and double-click it. Claude Desktop shows an install dialog.

Option B: command line. First quit Claude Desktop completely (macOS: ⌘Q; Windows: tray icon → Quit). Claude rewrites its config file from memory when it quits, which would drop the new entry. Then run:

bash
npx -y sysdesign-board install

Reopen Claude Desktop.

Option C: edit the config yourself (again, with Claude Desktop quit). The file is at:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
  • Linux (unofficial builds): ~/.config/Claude/claude_desktop_config.json
json
{  "mcpServers": {    "sysdesign-board": { "command": "npx", "args": ["-y", "sysdesign-board"] }  }}

On Windows, use "command": "cmd", "args": ["/c", "npx", "-y", "sysdesign-board"]. If Claude can't find npx (common with nvm, Volta or asdf), use the absolute path from which npx (macOS/Linux) or where npx (Windows).

Claude Code

Either add the server directly:

bash
claude mcp add sysdesign-board -- npx -y sysdesign-board

Or install the plugin, which also adds an interview skill, from this repo's marketplace. Run these inside Claude Code:

/plugin marketplace add tGhattas/sysdesign-board/plugin install sysdesign-board@sysdesign-board

Codex CLI

Add this to ~/.codex/config.toml (Windows: %USERPROFILE%\.codex\config.toml):

toml
[mcp_servers.sysdesign-board]command = "npx"args = ["-y", "sysdesign-board"]tool_timeout_sec = 120

ChatGPT app

ChatGPT connects to MCP servers over public HTTPS, so you run the board locally and expose only its /mcp endpoint through a tunnel. The board itself, its API and its WebSocket refuse any request that doesn't come from localhost.

  1. Start the board, and keep this running:
    bash
    npx -y sysdesign-board serve
  2. In another terminal, open a tunnel (here with cloudflared):
    bash
    cloudflared tunnel --url http://localhost:4321
  3. Get your private connector token:
    bash
    npx -y sysdesign-board token
  4. In ChatGPT, go to Settings → Apps & Connectors → Advanced, enable Developer Mode, then Create a connector with the URL https://<your-tunnel>.trycloudflare.com/mcp?token=<token> and no authentication.

Treat that URL like a password. To change the token, delete ~/.sysdesign-board/token, or set SYSDESIGN_TOKEN yourself.

Any other MCP client

json
{ "command": "npx", "args": ["-y", "sysdesign-board"] }

Use it

Start a new chat and say:

Start a system design interview with me on my sketch board.

Or use the interview prompt from your app's prompt or slash menu. The board opens at http://localhost:4321.

  • Draw your design. Talk with the 🎤 mic (your message is sent when you pause) or type in the panel.
  • The interviewer keeps listening between replies. It is woken by what you say, and when you stop drawing for a few seconds.
  • Click I'm done when you're finished to get a feedback card with scores.

The board also works without any AI: run npx -y sysdesign-board serve and open http://localhost:4321.

Tools the interviewer gets

ToolWhat it does
start_interviewSets the problem, level and duration, opens the board, and speaks the opening line
wait_for_candidateWaits for your speech or messages, or for board changes once you stop drawing
saySends a chat message, read aloud
get_boardStructured diagram: components, connections (with labels; dashed means async), zones, notes, and what changed
get_board_imageA PNG of the whole board (for handwriting and sketches)
annotate / clear_annotationsPin or remove "INTERVIEWER" comment bubbles on components
highlightMakes components glow, optionally zooming to them
draw_on_boardAdds components, connections, notes and zones in the same sketchy style
set_phaseRequirements → Estimation → API & Data → High-level → Deep dive → Wrap-up
end_interviewFeedback card with 1–5 scores, strengths and improvements
get_transcript, open_boardHousekeeping

Voice

  • Speech-to-text uses the browser's Web Speech API. Use Chrome, Edge or Safari; Firefox doesn't support it.
  • Text-to-speech uses your system voices. Toggle it with "Voice replies" in the panel. The mic pauses while the interviewer speaks, so it doesn't transcribe itself.
  • The native voice modes inside the ChatGPT and Claude apps can't be piped into another page. Those voice models are only available through paid APIs.

Keyboard

V select · H hand / hold Space · R rect · D diamond · O ellipse · A arrow · L line · P pen · T text · N sticky note · F zone · E eraser · K laser · C components · I interview panel · Q tool lock · G grid snap · ⌘/Ctrl Z undo · ⌘/Ctrl D duplicate · ⇧1 fit · ? all shortcuts

Privacy Policy

The full policy is in PRIVACY.md. In short:

  • What it handles: your board, the interview chat and voice transcripts, your settings, and a local access token.
  • Where it's stored: only on your machine, in the browser's local storage for the board page and in ~/.sysdesign-board/. The author runs no server, and there are no accounts, analytics or telemetry.
  • What leaves your computer:
    • When the AI interviewer calls a tool, the AI app you connected receives the board description, a picture of the board, your messages and your transcripts, under that provider's privacy policy.
    • The mic uses your browser's speech recognition. Chrome sends the audio to Google; Safari uses Apple's speech service.
    • npx downloads the package from npm.
    • With the optional tunnel, /mcp requests pass through the tunnel provider.
  • Security:
    • The board, its API and its WebSocket only answer localhost requests from localhost pages. Other websites and tunnels are refused, which also blocks DNS-rebinding attacks.
    • The HTTP endpoint (/mcp) always requires the token.
  • Sharing and retention: nothing is sold or shared. Data stays until you clear the board or delete ~/.sysdesign-board/.
  • Contact: open an issue.

Troubleshooting

  • The server doesn't appear in Claude Desktop. Claude was probably running when the config was edited and saved over it. Quit it fully, run npx -y sysdesign-board install, then reopen.
  • "npx: command not found" in Claude's logs. Use an absolute path to npx (see Option C), or run npx -y sysdesign-board install, which writes absolute paths for you.
  • Port 4321 is taken. Set SYSDESIGN_PORT (or --port=5000). In the .mcpb extension it's a setting.
  • The mic does nothing. Open the board in Chrome, Edge or Safari and allow microphone access.
  • Several apps connected at once? That's fine. The first one hosts the board and the others share it.

CLI

sysdesign-board            MCP server over stdio (what AI apps launch)sysdesign-board serve      Board + HTTP MCP endpoint onlysysdesign-board install    Add to Claude Desktop (quit it first)  [--npx | --local]sysdesign-board token      Print the HTTP MCP token / URL

Development

bash
git clone <this repo> && cd sysdesign-boardnpm installnpm run dev

npm run dev starts the hub on :4321 with reload, and Vite on :5173. Other scripts:

CommandWhat it does
npm run buildBuilds the board UI and bundles the server into one file
npm run build:mcpbPacks build/sysdesign-board-<version>.mcpb
npm run typecheckType-checks the web and server code
node scripts/mcp-smoke.mjsDrives every MCP tool against a running board
npm run lock:pluginRefreshes the Claude plugin's package-lock.json (run after the new version is on npm)

Releasing a new version:

  1. Run npm version <x.y.z> --no-git-tag-version. This also updates server.json and the plugin's version and pinned dependency.
  2. Run npm publish.
  3. Run npm run lock:plugin. The lockfile needs the published package's checksum, so this comes after publishing.
  4. Commit, then tag and push.
  5. Make a GitHub release with npm run build:mcpb.
  6. Run mcp-publisher publish for the MCP Registry.

The code lives in three folders:

  • web/: the board (React UI plus the canvas engine in web/src/engine.ts; renderer and component glyphs in web/src/render/)
  • server/: the hub (HTTP, WebSocket and persistence), the MCP tools and the CLI
  • shared/: the data model, component catalog and geometry

License

MIT

来源:README.md,提交 2b61d36

工具

13

open_board

Open the board

Open the sketch board in the user’s browser and return its URL.

无需输入参数。

start_interview

Start a system design interview

Start a new interview session on the board: sets the problem statement, timer and phase, opens the board, and posts your opening message (spoken aloud to the candidate).

输入参数leveltitle必填problem必填clear_boardopening_messageduration_minutes

wait_for_candidate

Listen to the candidate

Block until the candidate says something (typed or by voice), the board changes and then settles, or the timeout passes. Returns what they said, what changed on the board, and interview timing. Call this in a loop between your replies.

输入参数timeout_seconds

say

Speak to the candidate

Send a message to the candidate. It appears in the board chat and is read aloud. Keep it short and conversational (no markdown).

输入参数text必填speak

get_board

Read the board

Structured description of the board: components (type, label, id, position, zone), connections (from -> to, label, dashed=async), zones, notes and text, freehand strokes, plus what changed since you last looked.

输入参数include_json

get_board_image

Look at the board

A PNG picture of the whole board as the candidate sees it (handwriting, sketches, layout). Use when get_board mentions freehand strokes or you need visual context.

无需输入参数。

annotate

Pin a comment on the board

Pin an interviewer comment bubble on the board, pointing at a component (by id or label) or at a position. Great for "what if this fails?" style probes.

输入参数xytext必填target

highlight

Point at components

Make components glow on the candidate’s board (e.g. while asking about them). Optionally scroll/zoom their view to them.

输入参数focussecondstargets必填

draw_on_board

Draw on the board

Add components, connections, sticky notes and zones in the same sketchy style (marked as drawn by the interviewer). Use sparingly — the candidate should drive. Component kinds: user (User), browser (Web App), mobile (Mobile App), dns (DNS), cdn (CDN), lb (Load Balancer), gateway (API Gateway), firewall (Rate Limiter), server (App Server), service (Service), worker (Worker), lambda (Function), cron (Scheduler), sql (SQL DB), nosql (NoSQL DB), cache (Cache), storage (Object Store), search (Search Index), warehouse (Data Warehouse), queue (Queue), stream (Event Stream), pubsub (Pub/Sub), notify (Notifications), auth (Auth), monitor (Monitoring), ml (ML Model), external (External API). Connections reference a component by its ref/label (new or existing) or id. Positions are optional (auto-laid out to the right of existing content).

输入参数noteszonescomponentsconnections

clear_annotations

Remove interviewer comments

Remove your pinned comments (all, or specific ids).

输入参数ids

set_phase

Move to an interview phase

Update the phase tracker the candidate sees: requirements (Requirements), estimation (Estimation), api (API & Data), high-level (High-level), deep-dive (Deep dive), wrap-up (Wrap-up).

输入参数phase必填

end_interview

End the interview with feedback

Close the session and show a feedback card on the board. Scores are 1-5.

输入参数scores必填signalsummary必填strengths必填improvements必填

get_transcript

Conversation transcript

Full chat/voice transcript of the current session.

无需输入参数。

版本历史

1
  1. v0.1.2最新Oct 21, 2026

评分

已满足适用检查项的 90%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分
  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具必需15 分
  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分