TestFlight Feedback MCP Server
Enables AI assistants to access TestFlight beta tester feedback, including screenshots, crash logs, and text comments from App Store Connect. It works across any platform without requiring Xcode, using official API keys and optional browser automation for full text feedback.
README
TestFlight Feedback MCP Server
An MCP (Model Context Protocol) server that gives AI assistants access to TestFlight beta tester feedback from App Store Connect. No Xcode required.
Why?
Apple doesn't expose TestFlight feedback text/comments via their public API (fastlane discussion). This MCP server provides:
- Official API access to screenshot & crash feedback submissions (JWT auth)
- Optional browser automation for full text feedback via Apple's internal iris API (Playwright)
- Works on any platform — Windows, Linux, macOS — no Xcode needed
Tools
| Tool | Description |
|---|---|
list_apps |
List all apps in your App Store Connect account |
list_builds |
List TestFlight builds (filter by version, status) |
list_beta_testers |
List beta testers (filter by app, group, email) |
list_beta_groups |
List beta tester groups for an app |
list_feedback |
List all feedback: screenshots, crashes, and text comments |
get_feedback_detail |
Get full details for a specific feedback submission |
get_crash_log |
Download crash log content for a crash submission |
respond_to_feedback |
Email a tester that their feedback has been addressed |
Setup
1. Get App Store Connect API Keys
- Go to App Store Connect → Users and Access → Keys
- Click Generate API Key (requires Admin role)
- Download the
.p8file (you can only download it once) - Note the Key ID and Issuer ID
2. Install
cd testflight-feedback-mcp
npm install
npm run build
3. Configure
Set environment variables or create a .env file:
ASC_KEY_ID=YOUR_KEY_ID
ASC_ISSUER_ID=YOUR_ISSUER_ID
ASC_PRIVATE_KEY_PATH=./AuthKey_XXXXXXXX.p8
Or provide the key inline:
ASC_PRIVATE_KEY="-----BEGIN PRIVATE KEY-----\nMIGT...your key...\n-----END PRIVATE KEY-----"
4. Connect to Your AI Assistant
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"testflight-feedback": {
"command": "node",
"args": ["/absolute/path/to/testflight-feedback-mcp/dist/index.js"],
"env": {
"ASC_KEY_ID": "YOUR_KEY_ID",
"ASC_ISSUER_ID": "YOUR_ISSUER_ID",
"ASC_PRIVATE_KEY_PATH": "/path/to/AuthKey.p8"
}
}
}
}
Cursor
Add to .cursor/mcp.json:
{
"mcpServers": {
"testflight-feedback": {
"command": "node",
"args": ["/absolute/path/to/testflight-feedback-mcp/dist/index.js"],
"env": {
"ASC_KEY_ID": "YOUR_KEY_ID",
"ASC_ISSUER_ID": "YOUR_ISSUER_ID",
"ASC_PRIVATE_KEY_PATH": "/path/to/AuthKey.p8"
}
}
}
}
VS Code + GitHub Copilot
Add to .vscode/mcp.json:
{
"servers": {
"testflight-feedback": {
"command": "node",
"args": ["/absolute/path/to/testflight-feedback-mcp/dist/index.js"],
"env": {
"ASC_KEY_ID": "YOUR_KEY_ID",
"ASC_ISSUER_ID": "YOUR_ISSUER_ID",
"ASC_PRIVATE_KEY_PATH": "/path/to/AuthKey.p8"
}
}
}
}
Optional: Respond to Testers via Email (SMTP)
The respond_to_feedback tool looks up the tester's email from a feedback submission and sends them a response. Configure any SMTP provider:
SMTP_HOST=smtp.gmail.com
SMTP_PORT=587
SMTP_USER=your-email@gmail.com
SMTP_PASS=your-app-password # Use App Password for Gmail
SMTP_FROM=your-email@gmail.com # Optional, defaults to SMTP_USER
APP_NAME=My App # Shows in the email footer
Works with Gmail, Outlook, custom SMTP, etc. The tool:
- Fetches the submission to get the tester's email + build version
- Sends a nicely formatted HTML email with your message
- Returns confirmation with tester name, email, and message ID
Optional: Full Text Feedback (Browser Auth)
Apple's public API only returns screenshot and crash submissions — not text comments. To access text feedback, enable Playwright-based browser authentication:
npm install playwright
npx playwright install chromium
# Add to your env:
ENABLE_BROWSER_AUTH=true
ASC_USERNAME=your@apple.id
ASC_PASSWORD=your-password
Note: This uses Apple's internal
iris/v1/betaFeedbacksAPI. You may be prompted for 2FA on your Apple device. Session cookies are cached in memory for the server lifetime.
Example Usage
Once connected, ask your AI assistant:
- "Show me the latest TestFlight feedback for my app"
- "List all crash reports from the last build"
- "Get the crash log for submission XYZ"
- "Who are my beta testers?"
- "Show feedback from build 1.2.0"
Architecture
src/
├── index.ts # MCP server entry point, tool registration
├── auth/
│ ├── jwt.ts # JWT token generation (ES256, .p8 key)
│ └── browser.ts # Optional Playwright session for iris API
├── api/
│ ├── client.ts # HTTP client with JWT auth & pagination
│ ├── types.ts # TypeScript types for API responses
│ ├── apps.ts # /v1/apps endpoints
│ ├── builds.ts # /v1/builds endpoints
│ ├── testers.ts # /v1/betaTesters, /v1/betaGroups
│ └── feedback.ts # Screenshot/crash/iris feedback endpoints
└── tools/
├── list-apps.ts
├── list-builds.ts
├── list-testers.ts
├── list-feedback.ts
├── get-feedback-detail.ts
└── get-crash-log.ts
API Coverage
| Endpoint | Auth | Status |
|---|---|---|
/v1/apps |
JWT | ✅ |
/v1/builds |
JWT | ✅ |
/v1/betaTesters |
JWT | ✅ |
/v1/betaGroups |
JWT | ✅ |
/v1/betaFeedbackScreenshotSubmissions |
JWT | ✅ |
/v1/betaFeedbackCrashSubmissions |
JWT | ✅ |
/iris/v1/betaFeedbacks (text comments) |
Browser | ✅ (optional) |
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。