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.
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.

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:

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.

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
flowchartand it becomes an editable drawing; or turn what you drew into Mermaid to paste into a README. - Drag and drop. Drop an
.excalidrawfile 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 info → Run 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.
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.