Lousa

Lousa

Enables AI agents to draw animated, hand-drawn diagrams on a live whiteboard through MCP tools, allowing users to visually explain concepts and edit scenes interactively.

Category
访问服务器

README

Lousa

A whiteboard where your AI agent draws its explanations.

Your agent drawing a rate limiter explanation, stroke by stroke

Ask your agent "explain this visually" and it sketches an animated, hand-drawn diagram on a live board — instead of another wall of text. Lousa is an MCP server + web viewer: the agent authors a declarative scene (boxes, arrows, cylinders, notes), and the board renders it stroke by stroke, as if someone were drawing it in front of you.

Lousa is Portuguese for the classic school blackboard.

Why it's different

  • The agent is the artist. No prompt box, no OCR — your coding agent already has the context and draws through typed MCP tools.
  • Live, incremental drawing. Follow-ups edit the scene; the open tab animates only what changed, like watching the diagram evolve.
  • Point at things. Click elements in the viewer to select them, then tell your agent "improve what I selected" — it reads your selection through get_selection and targets exactly that.
  • Boards expire by default (12h, sliding) — a whiteboard, not a document graveyard. ttlHours: 0 keeps a board forever.

Quick start

Requires Node.js 18.17+.

# Claude Code
claude mcp add lousa -- npx -y lousa mcp

# or run the server directly
npx lousa            # viewer at http://127.0.0.1:4666

Any MCP host works — for stdio configuration use command npx, args ["-y", "lousa", "mcp"]. The stdio endpoint auto-starts the local server, so the viewer is always available. Then ask your agent to "draw how our auth flow works" and open the URL it returns.

MCP tools

Tool What it does
create_board Creates a board from a scene, returns {id, url}
update_board Replaces the scene; unchanged elements stay byte-identical so only changes animate
get_board / list_boards Read boards
get_selection What the user selected in the viewer (ids + full elements)
delete_board Deletes a board

The scene is a JSON array of elements — box, ellipse, diamond, cylinder, arrow (connects ids), line, text, note — rendered with rough.js. The full schema ships inside the tool definitions; agents need no extra setup.

Self-hosting a shared server

Run lousa on any Node host and point machines at it:

LOUSA_TOKEN=<secret> HOST=0.0.0.0 npx lousa        # server
LOUSA_URL=https://your-host LOUSA_TOKEN=<secret> npx lousa mcp   # each client

LOUSA_TOKEN gates board listing/creation/deletion and /mcp; per-board URLs act as capability links. AWS Lambda + S3 deployment is supported out of the box (lambda.js handler + LOUSA_BUCKET for storage); any HTTPS front (API Gateway, etc.) works.

Env var Default Purpose
PORT / HOST 4666 / 127.0.0.1 server bind
LOUSA_DATA_DIR ~/.lousa/boards filesystem board storage
LOUSA_BUCKET — use S3 storage instead of filesystem
LOUSA_TOKEN — require bearer token (header or ?t=) on global routes and MCP
LOUSA_URL — lousa mcp proxies to this remote server
LOUSA_DEFAULT_TTL_HOURS 12 default board expiry

Development

npm install
npm test          # node --test
node cli.js       # local server

License

AGPL-3.0 © Henrique Pires

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
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
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
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
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
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
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
mcp-server-qdrant

mcp-server-qdrant

这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。

官方
精选
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选