
Scenic MCP
A Model Context Protocol server that enables AI-driven automation and testing of Scenic Elixir applications through a TCP connection.
README
Scenic MCP
MCP (Model Context Protocol) server for Scenic Elixir applications. This enables AI-driven automation and testing of any Scenic app.
Architecture
This project consists of two parts:
- Elixir Library - A TCP server that runs inside your Scenic app
- MCP Server - A TypeScript/Node.js server that implements the MCP protocol
Your Scenic App (Elixir)
↓ adds dependency
scenic_mcp (Elixir TCP Server on port 9999)
↓ TCP connection
scenic-mcp (TypeScript MCP Server)
↓ MCP protocol
AI Assistant (Claude, Cline, etc.)
Quick Start
1. Add to your Scenic app
Add scenic_mcp
to your dependencies in mix.exs
:
defp deps do
[
{:scenic, "~> 0.11"},
{:scenic_mcp, path: "../scenic_mcp"} # or from hex/git when published
]
end
The TCP server will automatically start when your app starts.
2. Install the MCP server
For development (from this repo):
cd scenic_mcp
npm install
npm run build
For end users (when published):
npm install -g @scenic/mcp-server
3. Configure your MCP client
Add to your MCP client configuration (e.g., Claude Desktop, Cline):
{
"mcpServers": {
"scenic": {
"command": "node",
"args": ["/path/to/scenic_mcp/dist/index.js"]
}
}
}
Or if installed globally:
{
"mcpServers": {
"scenic": {
"command": "scenic-mcp"
}
}
}
Testing
Test the Elixir TCP server:
# Start your Scenic app with scenic_mcp
cd your_scenic_app
iex -S mix
# In another terminal, test the TCP connection
echo "hello" | nc localhost 9999
You should see a JSON response with Elixir system info.
Test the MCP integration:
- Start your Scenic app
- Make sure the MCP server is configured in your client
- Use the
hello_scenic
tool to test the connection
Development
Project Structure
scenic_mcp/
├── lib/ # Elixir source
│ ├── scenic_mcp.ex
│ └── scenic_mcp/
│ ├── application.ex
│ └── server.ex
├── src/ # TypeScript source
│ └── index.ts
├── dist/ # Compiled JavaScript
├── mix.exs # Elixir package
└── package.json # Node package
Building
# Install Elixir deps
mix deps.get
# Install Node deps
npm install
# Build TypeScript
npm run build
# Bundle for local dev
npm run bundle
Current Features
- ✅ Basic TCP server in Elixir
- ✅ MCP server in TypeScript
- ✅ Hello world communication test
Roadmap
- [ ] Viewport discovery and management
- [ ] Input injection (click, type, etc.)
- [ ] Screenshot capture
- [ ] Scene navigation
- [ ] Element inspection
- [ ] Custom app-specific tools
License
MIT
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。