AppsFlyer MCP Server

A Model Context Protocol (MCP) server for integrating AppsFlyer analytics data with AI assistants.

Features

• Fetch aggregate data reports from AppsFlyer Pull API
• Support for multiple report types: partners_report, partners_by_date_report, daily_report, geo_report, geo_by_date_report
• Secure API token authentication
• Type-safe input validation with Pydantic

Installation

git clone https://github.com/ysntony/appsflyer-mcp
cd appsflyer-mcp
uv sync

Configuration

Set up your AppsFlyer API credentials as environment variables:

export APPSFLYER_API_BASE_URL="https://hq1.appsflyer.com"
export APPSFLYER_TOKEN="your_api_token_here"

Or create a .env file:

APPSFLYER_API_BASE_URL=https://hq1.appsflyer.com
APPSFLYER_TOKEN=your_api_token_here

Usage

Running the MCP Server

uv run python run_server.py

MCP Configuration

Add to your MCP configuration file:

{
  "mcpServers": {
    "appsflyer": {
      "command": "uv",
      "args": ["run", "python", "run_server.py"],
      "cwd": "/path/to/appsflyer-mcp",
      "env": {
        "APPSFLYER_API_BASE_URL": "https://hq1.appsflyer.com",
        "APPSFLYER_TOKEN": "your_api_token_here"
      }
    }
  }
}

Available Tools

• get_aggregate_data: Fetch aggregate data reports from AppsFlyer Pull API
• test_appsflyer_connection: Test the connection to AppsFlyer API

Report Types

• partners_report: Partner performance data
• partners_by_date_report: Daily partner performance data
• daily_report: Daily aggregate data (default)
• geo_report: Geographic performance data
• geo_by_date_report: Daily geographic performance data

Development

uv sync --dev
pytest

License

MIT License