LetDraw MCP
Connect AI assistants and scripts to LetDraw to read and author diagrams via MCP or REST, including creating diagrams from code, exporting to Mermaid/D2, and sharing read-only links.
README
<p align="center"> <a href="https://letdraw.com"> <picture> <source media="(prefers-color-scheme: dark)" srcset="docs/assets/letdraw-logo-dark.svg"> <img src="docs/assets/letdraw-logo.svg" alt="LetDraw" width="320"> </picture> </a> </p>
<h1 align="center">LetDraw MCP + REST</h1>
<p align="center"> Connect AI assistants and your own scripts to <a href="https://letdraw.com">LetDraw</a> — read and author diagrams in your account through the <b>Model Context Protocol (MCP)</b> or the <b>public REST API</b>. </p>
<p align="center"> <a href="LICENSE"><img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-blue.svg"></a> <img alt="Node 18+" src="https://img.shields.io/badge/node-18%2B-brightgreen"> <img alt="MCP: stdio + HTTP" src="https://img.shields.io/badge/MCP-stdio%20%2B%20HTTP-8A2BE2"> </p>
LetDraw is a hand-drawn–style online whiteboard and diagramming app. With this integration an MCP client (Claude, Cursor, VS Code, Codex, …) or any HTTP client can list workspaces, create and update diagrams, build a diagram from code (compose / Kubernetes / DOT / Terraform / SQL / …), stamp real product icons, export to Mermaid or D2, and share read-only links.
This repository ships the client-side integration: a zero-config MCP bridge, a typed REST SDK, the API spec, and setup docs. The LetDraw app and server are not part of this repo.
What's here
| Path | What it is |
|---|---|
packages/mcp |
@letdraw/mcp — zero-config stdio bridge to the hosted MCP server (npx @letdraw/mcp) |
packages/sdk |
@letdraw/sdk — typed REST client for scripting |
spec/mcp-tools.md |
The MCP tools, resources, and scopes |
spec/openapi.yaml |
OpenAPI 3.1 spec for the REST API |
spec/capabilities.md |
"How to draw well in LetDraw" guide for AI clients |
docs/ |
Getting started + per-client setup (Claude, Cursor, VS Code, Codex) |
examples/ |
Copy-paste recipes |
Endpoints
REST base https://api.letdraw.com/api-v1
MCP URL https://api.letdraw.com/mcp
Both authenticate with a personal API token (Authorization: Bearer ld_live_…), not a login session. Create one in the LetDraw app under
Account → Developer. The exact URLs and a ready-to-paste client config are
shown there too.
60-second start (MCP)
The hosted server speaks Streamable-HTTP, so most clients connect directly. For a stdio client, or for the simplest possible setup, use the bridge:
// Claude Desktop — claude_desktop_config.json
{
"mcpServers": {
"letdraw": {
"command": "npx",
"args": ["-y", "@letdraw/mcp"],
"env": { "LETDRAW_TOKEN": "ld_live_YOUR_TOKEN" }
}
}
}
See docs/clients for Cursor, VS Code, Claude Code and Codex.
60-second start (REST)
curl https://api.letdraw.com/api-v1/workspaces \
-H "Authorization: Bearer ld_live_YOUR_TOKEN"
Or with the SDK:
import { LetdrawClient } from '@letdraw/sdk';
const ld = new LetdrawClient({ token: process.env.LETDRAW_TOKEN! });
const { workspaces } = await ld.listWorkspaces();
Scopes
Tokens carry scopes; each tool/route requires one:
diagrams:read— list/get workspaces, folders, diagrams, iconsdiagrams:write— create/update/delete/share diagrams, from-codeexport— export to Mermaid / D2ai:generate— generate a diagram from a prompt (uses your own AI key)
License
MIT — see LICENSE. "LetDraw" and the LetDraw logo are trademarks of their owner; see NOTICE.md.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。