Escalidrau MCP Server

Escalidrau MCP Server

MCP server for Escalidrau, a collaborative macOS whiteboard. Lets AI agents read, draw, edit, rearrange, and export diagrams on a shared canvas in real time.

Category
访问服务器

README

Escalidrau

A whiteboard for thinking out loud — with your AI and with your people, on the same canvas.

Escalidrau is a desktop app for macOS, Windows and Linux where you sketch diagrams by hand while your local AI agent draws right along with you: ask for the architecture of a system and it appears on your screen as it goes, fix it together, and when you want to show it to someone, invite them into a live jam.

The Escalidrau canvas with a serverless architecture diagram

Install

macOS (Apple Silicon or Intel) — paste in a terminal:

curl -fL https://github.com/aguara-guazu/escalidrau/releases/latest/download/Escalidrau-$(uname -m | sed s/x86_64/x64/).dmg -o /tmp/Escalidrau.dmg && open /tmp/Escalidrau.dmg

Drag Escalidrau to Applications and open it. From then on it updates itself.

Windows — paste in PowerShell:

irm https://github.com/aguara-guazu/escalidrau/releases/latest/download/Escalidrau-Setup-x64.exe -OutFile "$env:TEMP\Escalidrau-Setup.exe"; & "$env:TEMP\Escalidrau-Setup.exe"

The installer needs no admin rights and puts Escalidrau in your Start menu. On ARM devices swap x64 for arm64.

Linux — paste in a terminal:

curl -fL https://github.com/aguara-guazu/escalidrau/releases/latest/download/Escalidrau-$(uname -m).AppImage -o ~/Escalidrau.AppImage && chmod +x ~/Escalidrau.AppImage && ~/Escalidrau.AppImage

That is the portable build — nothing to install. If you prefer a package, Debian and Ubuntu users can grab the .deb from the releases page and install it with sudo apt install ./Escalidrau-amd64.deb.

Draw with your agent

You connect your agent once, from the app's MCP menu: one click for Claude Code, Claude Desktop or Codex. Nothing else to install.

After that it is all conversation:

  • "Draw the checkout flow with API Gateway, Lambda and DynamoDB" → it appears on your canvas as it is written.
  • "Separate the diagrams that overlap and lay everything out horizontally" → each diagram moves as a whole, arrows and labels included.
  • "Export it as a PNG to my desktop" → ready to drop into a doc or a ticket.

Your agent looks at what it drew: it inspects the canvas and fixes what came out wrong — text overflowing a shape, crossed arrows, overlapping parts — before telling you it is done. It also sees what you do: move a box or change a label and it works from there.

Installed icon packs? It uses them. With the AWS pack, an AWS diagram comes out with the real icons:

Exported diagram using AWS icons

Jams: drawing with other people

Hit Start group jam, get a code, share it. Up to 10 people on the same canvas: you see each other's cursors live, each with their name and colour, and whatever anyone draws shows up on everyone's screen.

A jam with three people collaborating on the same diagram

While the jam is live, the button turns into the room code plus a Leave jam button.

The nice part of the model: the drawing travels straight between the computers (peer to peer, like the multiplayer in a shooter), without passing through any server of ours. Which means:

  • Nothing is stored anywhere. What you draw exists only in the apps of the people connected — whoever wants to keep it saves it on their machine (the app asks when you close it).
  • The room does not depend on whoever created it. If the host leaves, everyone else keeps drawing and someone takes over automatically.
  • Everyone keeps their own agent. It can be three people and three AIs on the same diagram.

What else is in the box

  • Mermaid both ways. Paste a flowchart and it becomes an editable drawing; or turn what you drew into Mermaid to paste into a README.
  • Drag and drop. Drop an .excalidraw file or a Mermaid file on the window and it offers to import it.
  • Icon packs. Install any of them from the public catalog in one click; they stay installed.
  • It keeps itself current. On macOS it installs new versions on launch and tells you what changed; on Windows and Linux it lets you know when one is out.
  • Export to PNG, SVG, or a file you can keep editing later.

Quick answers

Do I need an account? No. No login, no servers, no telemetry.

macOS says the app "is damaged". That happens when you download the DMG with a browser or get it over chat, because the app is not signed by Apple. The install command above avoids it. If it already happened:

xattr -dr com.apple.quarantine "/Applications/Escalidrau.app"

Windows shows a SmartScreen warning. Same reason: the installer is not signed. Choose More infoRun anyway.

Which systems are supported? macOS (Apple Silicon and Intel), Windows (x64 and ARM) and Linux (x64 and ARM, AppImage or .deb).

Does it update itself everywhere? The macOS build installs updates on its own. The Windows and Linux builds tell you when a new version is out and take you to the download — replacing an installed app in place is riskier on those platforms, so it is not done behind your back.

Someone cannot join a jam. Some restrictive networks block direct computer-to-computer connections. That is the network, not the app.

For developers

While the app is open it exposes an MCP server at http://localhost:3580/mcp with these tools: get_scene, get_layout, get_library, view_library, add_library_item, add_elements, update_elements, move_elements, delete_elements, import_mermaid, export_mermaid, view_canvas, export_image, wait_for_user_changes.

npm install
npm run dev    # web (vite, :3579) + server (:3580)
npm run app    # Electron app in dev mode
npm run dist         # macOS build into desktop/release/
npm run dist:win     # Windows build (needs Wine when run from macOS or Linux)
npm run dist:linux   # Linux build (AppImage + deb)

Releases are built by CI: pushing a v* tag runs each platform on its own runner and attaches every artifact to the release.

License

MIT — see LICENSE, which includes attribution for the dependencies the app redistributes.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选