review10 min read6h ago

Excalidraw MCP Server Review 2026: Let Claude Draw Your Diagrams

The Excalidraw MCP server lets AI clients create and edit hand-drawn diagrams with smart layout, Mermaid conversion, live preview, and PNG/SVG export. Full review: implementations, install steps, and comparisons.

Excalidraw MCP Server Review 2026: Let Claude Draw Your Diagrams
excalidraw mcpexcalidrawdiagram mcpmcp servermermaidwhiteboardmodel context protocolclaude

TL;DR — Excalidraw MCP Server Review 2026

The Excalidraw MCP server lets an AI agent draw — it turns natural-language requests into editable Excalidraw diagrams with shapes, arrows, smart layout, Mermaid conversion, and PNG/SVG/JSON export, often with a live browser preview that updates over websockets. The official excalidraw/excalidraw-mcp is the most-starred option (3,400+ stars), with several capable community forks. It is free, open source, installs in one command, and works in Claude Desktop, Cursor, Windsurf, and Cline. If you want Claude to sketch architecture diagrams, flowcharts, and wireframes you can then hand-edit, this is the tool.

Rating: 4.4/5 · Best free diagram-as-code MCP · Reviewed on Skiln

Table of Contents

  1. What Is the Excalidraw MCP Server?
  2. Why Diagram-as-Code for AI Matters
  3. Which Excalidraw MCP Should You Use?
  4. Key Features
  5. How to Install the Excalidraw MCP
  6. Excalidraw MCP vs Mermaid vs Figma MCP
  7. What You Can Build
  8. Tips for Better AI-Generated Diagrams
  9. Limitations
  10. Verdict
  11. Frequently Asked Questions

What Is the Excalidraw MCP Server?

Excalidraw is the beloved open-source virtual whiteboard with the hand-drawn aesthetic — millions of people use it for architecture diagrams, flowcharts, and quick sketches. The Excalidraw MCP server connects that whiteboard to an AI client through the Model Context Protocol, so instead of dragging shapes yourself, you describe what you want and the agent draws it.

Tell Claude "diagram a three-tier web app with a load balancer, two app servers, and a Postgres primary with a read replica," and the Excalidraw MCP creates the boxes, labels, and arrows on a real canvas. Because the output is native Excalidraw elements — not a flattened image — you can then open the file and tweak anything by hand. That round-trip between AI generation and manual editing is the whole point.

Why Diagram-as-Code for AI Matters

LLMs are good at structure. They understand that a load balancer sits in front of app servers, that a flowchart has decision diamonds, that an ER diagram has cardinality. What they historically could not do is draw — they could emit Mermaid text, but not place and style elements on a canvas.

The Excalidraw MCP closes that gap. It gives the model real drawing tools: create element, update element, group, align, distribute, connect with a bound arrow. The result is that diagramming becomes conversational. You iterate by talking ("make the database cluster bigger, move auth above the gateway") instead of fiddling with a mouse. For engineers documenting systems, this is a genuine time-saver — and the hand-drawn Excalidraw style keeps the output looking like a thoughtful sketch rather than sterile boxes.

Which Excalidraw MCP Should You Use?

There is no single canonical package — "Excalidraw MCP" refers to a family of implementations. The ones worth knowing:

  • excalidraw/excalidraw-mcp — the official one from the Excalidraw team, billed as a "fast and streamable Excalidraw MCP App." Most stars (3,400+), the safest default.
  • cmd8/excalidraw-mcp — clean Cursor-focused setup, points at a specific .excalidraw file.
  • yctimlin/mcp_excalidraw — MCP server and a Claude Code skill, full element control with real-time canvas sync.
  • whallysson/excalidraw-mcp — adds a real-time collaborative frontend.
  • scofieldfree/excalidraw-mcp — one-line claude mcp add install.

Start with the official server unless you need a specific feature (Docker deploy, collaborative frontend, a bundled skill) that a fork provides. Skiln tracks stars and last-commit dates for each so you can pick the best-maintained one at any time.

Key Features

Across the leading implementations, the capabilities that matter:

  • Full element control — create, update, delete, and query any Excalidraw element (rectangles, ellipses, text, arrows, lines, frames).
  • Smart layout — automatically calculates text width and binds labels to their containers, so boxes size themselves correctly.
  • Real-time preview — a local browser window updates instantly over websockets as the AI works.
  • Mermaid support — convert Mermaid flowcharts and sequence diagrams into editable Excalidraw elements.
  • Export options — PNG, SVG, and native .excalidraw JSON.
  • Advanced manipulation — group, align, distribute, lock, and unlock elements.
  • Templates — some builds ship built-in architecture-diagram templates.
  • Docker support — containerized deployment in several forks.

How to Install the Excalidraw MCP

All implementations need Node.js 18+. The fastest install, Claude-first:

claude mcp add excalidraw -- npx -y @scofieldfree/excalidraw-mcp

For Cursor, add to ~/.cursor/mcp.json:

{   "mcpServers": {     "excalidraw": {       "command": "npx",       "args": ["-y", "@cmd8/excalidraw-mcp", "--diagram", "/path/to/diagram.excalidraw"]     }   } }

For the official server or a fork with a frontend, clone the repo, run pnpm install, then pnpm dev to start the websocket-backed preview, and point your MCP client at the server. Restart the client, confirm the Excalidraw tools appear, and ask the agent to draw something simple as a smoke test. The Skiln Config Generator can fold Excalidraw into a multi-server config.

Excalidraw MCP vs Mermaid vs Figma MCP

Three ways an AI can produce visuals, compared:

ServerBest ForOutputSweet SpotCost
Excalidraw MCPHand-drawn diagrams + whiteboardingEditable .excalidraw, PNG, SVGArchitecture, flowcharts, sketchesFree
Mermaid MCPText-defined diagramsRendered image / Mermaid codeDocs-as-code, version-controlled diagramsFree
Figma MCPHigh-fidelity UI/UX designFigma files, design tokensProduction UI design, design systemsFree + Figma plan

Use Excalidraw MCP for fast, editable, good-looking system sketches. Use the Mermaid MCP when you want diagrams that live as text in your repo. Use the Figma context MCP when you are doing real product design and need pixel-level fidelity. Many teams keep all three. For the broader picture, see our best diagram and whiteboard MCP servers roundup.

What You Can Build

  • Architecture diagrams for design docs and RFCs, generated from a paragraph of description.
  • Flowcharts for onboarding, auth, and checkout flows, iterated by conversation.
  • ER diagrams from a schema dump, then hand-arranged for clarity.
  • Wireframes and low-fi UI sketches for early product discussions.
  • Mermaid migration — paste existing Mermaid and get editable Excalidraw you can restyle.
  • Slide visuals — export PNG/SVG straight into a deck.

Tips for Better AI-Generated Diagrams

  • Describe structure, not coordinates. Say "auth service sits between the gateway and the database," not "put a box at x=200." Let smart layout handle placement.
  • Iterate in small steps. Generate a skeleton, then refine ("add a cache between app and DB"). Big one-shot prompts produce messy canvases.
  • Ask for JSON export when you want to keep editing later, PNG/SVG when you just need to embed.
  • Use Mermaid as an intermediate for complex flows — generate Mermaid first, convert, then polish.
  • Keep the preview window open so you can catch a wrong layout early and correct it in the next message.

Limitations

  • Fragmented ecosystem. Multiple implementations with different feature sets; you have to pick one and learn its quirks.
  • Some are "work in progress." A few repos are explicitly early; check stars and recent commits before relying on one.
  • Not a design tool. For production UI fidelity, reach for Figma — Excalidraw is intentionally low-fi.
  • Complex diagrams still need a human pass. AI gets you 80% of the way; final alignment and labeling usually want a manual touch.

Verdict

The Excalidraw MCP server is one of the most satisfying MCP integrations to actually use — watching Claude draw a clean architecture diagram in a live preview window feels like the future of documentation. It is free, open source, installs in a minute, and produces editable output you own. The only friction is choosing among implementations; start with the official excalidraw/excalidraw-mcp and branch out only if you need a specific feature.

For engineers who document systems, write RFCs, or just think better with a picture, this earns a spot in your MCP stack. Pair it with the best documentation MCP servers for an AI-assisted docs workflow end to end.

Want AI that can draw? Browse every diagram, whiteboard, and visualization MCP server on Skiln, ranked by stars and active maintenance.

Browse Now →

Frequently Asked Questions

What is the Excalidraw MCP server?

The Excalidraw MCP server is a Model Context Protocol server that lets AI clients like Claude Desktop, Cursor, and Cline create and edit Excalidraw diagrams programmatically. The agent can add shapes, text, arrows, and containers, arrange them with smart layout, convert Mermaid syntax to Excalidraw, and export to PNG, SVG, or JSON — often with a live browser preview that updates over websockets.

Is there an official Excalidraw MCP server?

Yes. The Excalidraw team maintains excalidraw/excalidraw-mcp, described as a fast and streamable Excalidraw MCP app, and it is the most-starred implementation. Several strong community versions also exist (cmd8, whallysson, yctimlin/mcp_excalidraw, scofieldfree), each with slightly different features like Docker support or a collaborative frontend. Skiln indexes all of them so you can compare stars and maintenance.

How do I install the Excalidraw MCP in Claude?

Most implementations install with a single command. For example: claude mcp add excalidraw -- npx -y @scofieldfree/excalidraw-mcp. For Cursor, add an entry to ~/.cursor/mcp.json with command npx and the package name in args. You need Node.js 18 or newer. Restart the client and the Excalidraw tools appear in the MCP list.

Can the Excalidraw MCP convert Mermaid diagrams?

Several implementations can. They take Mermaid syntax — flowcharts, sequence diagrams, ER diagrams — and render it as native, editable Excalidraw elements rather than a flat image. This is useful when you have existing Mermaid in your docs and want to hand-tweak the result in Excalidraw's hand-drawn style.

Does the Excalidraw MCP show a live preview?

Yes, in the implementations that ship a frontend. They open a local browser window connected over websockets, so every change the AI makes appears instantly on the canvas. You watch the diagram build itself in real time, then export it or keep editing by hand in the normal Excalidraw editor.

What can I export from the Excalidraw MCP?

Common export formats are PNG (for embedding in docs and slides), SVG (for crisp, scalable vector output), and .excalidraw JSON (the native format you can reopen and edit later). The JSON export is the important one for round-tripping — it keeps every element editable instead of flattening to a picture.

Is the Excalidraw MCP free?

Yes. Excalidraw itself is open source, and every Excalidraw MCP implementation we track is free and open source on GitHub. You only pay for the underlying AI client usage. There is no separate license fee for the MCP server.

Where can I find other diagram and whiteboard MCP servers?

Skiln indexes diagram, whiteboard, and visualization MCP servers across every major registry. Read our roundup of the best diagram and whiteboard MCP servers for the full landscape, including Mermaid, Figma, and architecture-diagram tools, or browse the category directly at /mcps.


Last updated: June 24, 2026 · Skiln tracks new MCP servers daily across PulseMCP, Smithery, Glama, LobeHub, and mcp.directory.

Frequently Asked Questions

What is the Excalidraw MCP server?
The Excalidraw MCP server is a Model Context Protocol server that lets AI clients like Claude Desktop, Cursor, and Cline create and edit Excalidraw diagrams programmatically. The agent can add shapes, text, arrows, and containers, arrange them with smart layout, convert Mermaid syntax to Excalidraw, and export to PNG, SVG, or JSON — often with a live browser preview that updates over websockets.
Is there an official Excalidraw MCP server?
Yes. The Excalidraw team maintains excalidraw/excalidraw-mcp, described as a fast and streamable Excalidraw MCP app, and it is the most-starred implementation. Several strong community versions also exist (cmd8, whallysson, yctimlin/mcp_excalidraw, scofieldfree), each with slightly different features like Docker support or a collaborative frontend. Skiln indexes all of them so you can compare stars and maintenance.
How do I install the Excalidraw MCP in Claude?
Most implementations install with a single command. For example: claude mcp add excalidraw -- npx -y @scofieldfree/excalidraw-mcp. For Cursor, add an entry to ~/.cursor/mcp.json with command npx and the package name in args. You need Node.js 18 or newer. Restart the client and the Excalidraw tools appear in the MCP list.
Can the Excalidraw MCP convert Mermaid diagrams?
Several implementations can. They take Mermaid syntax — flowcharts, sequence diagrams, ER diagrams — and render it as native, editable Excalidraw elements rather than a flat image. This is useful when you have existing Mermaid in your docs and want to hand-tweak the result in Excalidraw's hand-drawn style.
Does the Excalidraw MCP show a live preview?
Yes, in the implementations that ship a frontend. They open a local browser window connected over websockets, so every change the AI makes appears instantly on the canvas. You watch the diagram build itself in real time, then export it or keep editing by hand in the normal Excalidraw editor.
What can I export from the Excalidraw MCP?
Common export formats are PNG (for embedding in docs and slides), SVG (for crisp, scalable vector output), and .excalidraw JSON (the native format you can reopen and edit later). The JSON export is the important one for round-tripping — it keeps every element editable instead of flattening to a picture.
Is the Excalidraw MCP free?
Yes. Excalidraw itself is open source, and every Excalidraw MCP implementation we track is free and open source on GitHub. You only pay for the underlying AI client usage. There is no separate license fee for the MCP server.
Where can I find other diagram and whiteboard MCP servers?
Skiln indexes diagram, whiteboard, and visualization MCP servers across every major registry. Read our roundup of the best diagram and whiteboard MCP servers for the full landscape, including Mermaid, Figma, and architecture-diagram tools, or browse the category directly at /mcps.

Stay in the Loop

Join 1,000+ developers. Get the best new Skills & MCPs weekly.

No spam. Unsubscribe anytime.