Shotter
Enables AI assistants to automate iOS Simulator interactions including device management, UI element interaction (tap, swipe, type), screenshot capture, and execution of YAML-defined navigation workflows.
README
Shotter
MCP (Model Context Protocol) server for iOS Simulator automation. Provides tools for AI assistants to navigate iOS apps, interact with UI elements, and capture screenshots.
Features
- MCP Integration: Works with Claude Desktop, Claude Code, and other MCP-compatible clients
- Device Management: List, select, and boot iOS Simulator devices
- UI Automation: Tap, swipe, type, and navigate iOS apps
- Screenshot Capture: Visual analysis and high-quality screenshot saving
- Workflow Support: Load and execute YAML-defined navigation workflows
Installation
npm install
npm run build
Configuration
Claude Desktop
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"shotter": {
"command": "node",
"args": ["/path/to/shotter/dist/server.js"]
}
}
}
Claude Code
Add to your project's .mcp.json or global settings:
{
"mcpServers": {
"shotter": {
"command": "node",
"args": ["/path/to/shotter/dist/server.js"]
}
}
}
Available Tools
Device Management
| Tool | Description |
|---|---|
list_devices |
List available iOS simulators with names, UDIDs, and states |
select_device |
Boot a simulator and set it as active (fuzzy name matching) |
launch_app |
Launch an app by bundle ID on the active device |
UI Interaction
| Tool | Description |
|---|---|
ui_view |
Capture screenshot for visual analysis |
ui_describe_all |
Get accessibility info with coordinates for all UI elements |
ui_tap |
Tap at specific coordinates |
ui_swipe |
Swipe gesture (for scrolling and navigation) |
ui_type |
Type text into focused field |
screenshot |
Save high-quality screenshot to a file |
Workflow
| Tool | Description |
|---|---|
load_workflow |
Load and parse a workflow YAML file |
list_workflows |
List available workflow files in a directory |
Available Prompts
| Prompt | Description |
|---|---|
navigate |
System prompt with iOS Simulator navigation strategies |
workflow-step |
Format a workflow step with goal and hints for execution |
Usage Examples
Once configured, ask your AI assistant to:
"Select iPhone 16 Pro and launch the Settings app"
"Take a screenshot of the current screen"
"Tap on the Wi-Fi option in Settings"
"Navigate to Privacy settings and take a screenshot"
Workflow Format
Workflows define multi-step navigation sequences:
name: "My App Screenshots"
description: "Capture key screens from my app"
bundleId: "com.example.myapp"
devices:
- "iPhone 16 Pro"
- "iPad Pro"
maxIterations: 20
outputDir: "./screenshots"
steps:
- goal: "Navigate to settings"
hints:
- "Look for a gear icon or Settings tab"
screenshot: "settings.png"
- goal: "Open the profile screen"
screenshot: "profile.png"
YAML Fields
| Field | Required | Description |
|---|---|---|
name |
Yes | Workflow name |
description |
No | Workflow description |
bundleId |
Yes | App bundle identifier |
devices |
Yes | List of simulator device names (fuzzy matched) |
runBefore |
No | Script to run before steps for each device |
maxIterations |
No | Max iterations per step (default: 20) |
outputDir |
No | Screenshot output directory (default: ./screenshots) |
steps |
Yes | List of navigation steps |
Step Fields
| Field | Required | Description |
|---|---|---|
goal |
Yes | What to accomplish |
hints |
No | Hints to help navigate |
screenshot |
No | Filename to save screenshot after step completes |
Requirements
- macOS with Xcode installed
- iOS Simulator
- Node.js 18+
- Facebook IDB (
brew install idb-companion)
Development
npm run dev # Watch mode
npm run build # Compile TypeScript
How It Works
Shotter acts as an MCP server that:
- Manages iOS Simulator device lifecycle (boot, select)
- Proxies UI commands to
ios-simulator-mcpfor screen interaction - Provides navigation prompts optimized for AI assistants
- Supports workflow files for repeatable multi-step automation
The AI assistant uses ui_view to see the screen, makes decisions about what to tap or type, and verifies actions succeeded before continuing.
License
ISC
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。