MCP.so
Sign In

mcp-excalidraw

@yctimlin

RepositoryHomepage

About mcp-excalidraw

MCP server and Claude Code skill for Excalidraw — programmatic canvas toolkit to create, edit, and export diagrams via AI agents with real-time canvas sync.

#mcp-excalidraw#excalidraw#drawing-api

Basic information

Category

Developer Tools

License

MIT

Runtime

node

Transports

stdio

Publisher

yctimlin

Submitted by

yctimlin

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mcp_excalidraw": {
      "command": "docker",
      "args": [
        "build",
        "-t",
        "mcp/excalidraw",
        "."
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is mcp-excalidraw?

mcp-excalidraw gives AI agents a live Excalidraw canvas they can draw on, inspect, refine, and save into your repo. It lets coding agents create architecture diagrams and flowcharts programmatically, see their work via screenshots, fix layout problems, and export .excalidraw files that can be committed alongside code. It works with three interfaces: a CLI/agent skill, an MCP server (26 tools over stdio), and a REST API.

How to use mcp-excalidraw?

Installation requires Node.js ≥ 18. For coding agents, run npx -y mcp-excalidraw-server install-skill --dir <skills-root> to install the portable skill, then use CLI commands like add, describe, screenshot, export, and import. For MCP clients, add the provided npx configuration to the client’s settings. The canvas server auto-starts on http://127.0.0.1:3000 when a drawing command is invoked (opt out with EXCALIDRAW_NO_AUTOSTART=1). Explicit start and stop commands manage the server.

Key features of mcp-excalidraw

  • 26 MCP tools for full element-level CRUD, layout, snapshots, and file I/O.
  • CLI-first – composable commands with JSON output and meaningful exit codes.
  • Auto‑starting canvas server – no separate setup for drawing commands.
  • Agent sees the canvas – describe (structured text) and screenshot (image) enable iterative refinement.
  • Mermaid conversion – mermaid / create_from_mermaid and share for encrypted URLs.
  • Multi‑agent support – multiple agents on the same canvas concurrently.

Use cases of mcp-excalidraw

  • Coding agents drawing architecture diagrams from a single prompt and iterating until labels fit and arrows route cleanly.
  • Flowcharts and diagrams as code artifacts – export .excalidraw files into the repo, commit, re‑import, and refine.
  • Self‑verifying diagrams – the agent screenshots its own work and fixes truncation or overlap before marking a task complete.
  • Diagrams in CI or headless environments – CLI works without a browser tab (except for screenshot and Mermaid features).

FAQ from mcp-excalidraw

What is the difference between mcp-excalidraw and the official Excalidraw MCP?

The official MCP is a one‑shot chat widget that streams a diagram inline. mcp-excalidraw gives coding agents a persistent, editable canvas with element‑level CRUD, screenshots, layout tools, snapshots, file export/import, and Mermaid conversion. It supports multi‑agent usage and works without MCP (via CLI + REST API).

What are the runtime requirements?

Node.js ≥ 18. No API keys are required – core drawing runs fully local and is MIT licensed. Mermaid conversion runs in the local browser canvas; the share feature optionally uploads an encrypted scene to excalidraw.com.

How does the canvas server start?

Canvas‑driving CLI commands and the MCP server auto‑start the canvas server if nothing is listening on http://127.0.0.1:3000. Set EXCALIDRAW_NO_AUTOSTART=1 to opt out. The start command overrides this opt‑out.

Is there authentication for the canvas server?

The canvas server binds only to 127.0.0.1 by default and has

Comments

More Developer Tools MCP servers

test

cloudwego

The ultimate LLM/AI application development framework in Go.

12.2K

Code Index MCP

johnhuang316

A Model Context Protocol (MCP) server that helps large language models index, search, and analyze code repositories with minimal setup

979

TalkToFigma

sonnylazuardi

TalkToFigma: MCP integration between AI Agent (Cursor, Claude Code, Codex) and Figma, allowing Agentic AI to communicate with Figma for reading designs and modifying them programmatically.

6.9K

MCP-Scan: An MCP Security Scanner

invariantlabs-ai

Security scanner for AI agents, MCP servers and agent skills.

2.8K

MCP Containers

metorial

Connect any AI model to 1200+ integrations (MCP, CLI, API)

3.3K