Guides

Integrate SDK with AI

Offload much of the AppsFlyer SDK integration work to AppsFlyer's SDK MCP server by using an AI-powered coding tool, such as OpenAI Codex, GitHub Copilot, or Cursor. This guide applies to both iOS and Android app projects.

Step 1: Get ready

  • Your AppsFlyer Dev key.
  • Your app project (iOS or Android).
  • Your app identifier:
    • Android: your app's package name (for example, com.example.myapp).
    • iOS: your app's bundle ID or App Store ID (for example, id123456789).
  • Your AI-powered coding tool is installed and set up for your project.
  • Your IDE is installed to sync, build, and run the app:
    • Android: Android Studio.
    • iOS: Xcode.

Step 2: Add the AppsFlyer SDK MCP server

The AppsFlyer SDK MCP server is the same for both platforms. The only values that change are your DEV_KEY and APP_ID environment variables. Choose the instructions for your AI coding tool below.

Cursor

  1. In Cursor, go to Cursor > Settings… > Cursor settings > Tools & MCP.

  2. Click New MCP server.

  3. Add a new MCP server entry in mcp.json, using your own app's APP_ID and DEV_KEY:

    "appsFlyer-sdk-mcp": {
  "command": "npx",
  "args": [
    "-y",
    "@appsflyer/sdk-mcp-server"
  ],
  "env": {
    "APP_ID": "<your-app-identifier>",
    "DEV_KEY": "<your-dev-key>"
  }
}

  4. Restart Cursor to apply the changes.

  5. In Cursor, go to Cursor > Settings… > Cursor settings > Tools & MCP.

  6. Find the MCP server entry you added.

  7. Confirm it's marked with a green dot. This indicates the MCP server is working.

Claude Code

  1. In your project terminal, run the following Claude Code command, using your own app's APP_ID and DEV_KEY:

    claude mcp add --transport stdio appsFlyer-sdk-mcp \
  -e APP_ID=<your-app-identifier> \
  -e DEV_KEY=<your-dev-key> \
  -- npx -y @appsflyer/sdk-mcp-server

  2. Restart Claude Code to load the new server.

  3. Confirm that appsFlyer-sdk-mcp is listed and connected (run claude mcp list).

Codex

  1. In Codex, go to Settings > MCP Servers.
  2. Click + Add server.
  3. In the Name box, give your server a name.
  4. In Command to launch, enter npx.
  5. In Arguments, enter the following arguments, each in a different field:
    • -y (to install and update your MCP each run).
    • @appsflyer/sdk-mcp-server (the server name as listed on the NPM registry).
  6. In Environment variables, add the variables DEV_KEY and APP_ID, and their values for your app.
  7. Restart Codex to load the new MCP server.
  8. In Codex, open Settings > MCP Servers and confirm that the AppsFlyer SDK MCP server is turned on.

GitHub Copilot (Android only)

Note: These GitHub Copilot steps apply to Android projects in Android Studio. For iOS, use Codex or Cursor above.

  1. In Android Studio, open Settings and go to Tools > GitHub Copilot > Model Context Protocol (MCP).

  2. In the Model Context Protocol (MCP) settings, click Configure.

  3. In mcp.json, add the AppsFlyer SDK MCP server settings, using your own app's APP_ID and DEV_KEY:

    "appsFlyer-sdk-mcp": {
  "command": "npx",
  "args": [
    "-y",
    "@appsflyer/sdk-mcp-server"
  ],
  "env": {
    "APP_ID": "<your-app-identifier>",
    "DEV_KEY": "<your-dev-key>"
  }
}

  4. Restart Android Studio to apply the changes.

  5. In Android Studio, open the GitHub Copilot Chat.

  6. At the bottom of the Copilot Chat panel, select Agent from the agents dropdown.

  7. Click the MCP Tools button.

  8. In the Configure Tools dialog, for the appsFlyer-sdk-mcp tool, click Start.

  9. To confirm that the server is running, open the GitHub Copilot MCP Log. In the displayed MCP log panel, you should see that the server is running.

Step 3: Prompt the AI to integrate the SDK

  1. In the chat, type Integrate the AppsFlyer SDK using MCP.
  2. Review the changes in your codebase in your editor, and verify that the integrated code matches your project requirements and coding standards. If anything looks unfamiliar, ask the AI follow-up questions.

Step 4: Prompt the AI to verify the integration

  1. Open your IDE (Android Studio for Android, Xcode for iOS).
  2. Sync the project.
  3. Build and run your app.
  4. In the chat, prompt the AI to verify the AppsFlyer SDK integration. The AI agent will use the MCP tool to read logs from your device and confirm whether the integration succeeded.
    • If errors are detected, the tool will explain what went wrong and suggest fixes.
    • If everything looks good, you'll see confirmation of a successful integration.