Applovin-Max-MCP-Server by FurkanEmiroglu - MCP Server

AppLovin Max MCP Server

A Model Context Protocol (MCP) server that provides LLM clients like Claude with access to AppLovin Max advertising platform analytics and reporting capabilities.

⚠️ IMPORTANT: PROOF OF CONCEPT ONLY

This is a proof-of-concept throwaway project created for demonstration and experimentation purposes. It is NOT production-ready and should NOT be used in production environments without significant additional work.

Known Limitations:

No comprehensive error handling or recovery mechanisms
No rate limiting or request throttling
No input validation beyond basic parameter checks
No logging or monitoring capabilities
No unit tests or integration tests
No security audits performed
No performance optimization or scalability considerations
API keys transmitted in query parameters (consider more secure methods for production)

Use at your own risk. This code is provided as-is with no warranties. See the LICENSE file for details.

Overview

This Go-based MCP server exposes AppLovin Max API functionality through standardized tools, enabling AI assistants to query mobile ad revenue data, analyze user cohorts, and generate insights from your AppLovin monetization metrics.

Features

Revenue Reporting: Access aggregated mediation statistics including impressions, revenue, eCPM, and fill rates
Cohort Analysis: Track user lifecycle metrics segmented by installation date
Flexible Filtering: Filter by application, country, platform, ad type, and more
Multiple Data Views: Choose between revenue, impression, or session-based cohort analysis
Time-Series Tracking: Monitor how metrics evolve over time (up to 45 days post-install)

Prerequisites

Go 1.25.4 or later
AppLovin Max account with API access
AppLovin API key

Installation

Clone the repository:

git clone <repository-url>
cd Applovin-Max-MCP

Install dependencies:

go mod download

Build the server:

go build -o bootstrap cmd/main.go

Configuration

Set your AppLovin API key as an environment variable:

export APPLOVIN_API_KEY="your_api_key_here"

Usage

Starting the Server

The MCP server communicates via stdio (standard input/output):

./bootstrap

Integrating with Claude Desktop

Add the following configuration to your Claude Desktop config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "applovin-max": {
      "command": "/path/to/Applovin-Max-MCP/bootstrap",
      "env": {
        "APPLOVIN_API_KEY": "your_api_key_here"
      }
    }
  }
}

Available Tools

1. Revenue Report (`revenue_report`)

Retrieves aggregated mediation statistics from AppLovin Max.

Required Parameters

start (string): Start date in YYYY-MM-DD format
end (string): End date in YYYY-MM-DD format
format (string): Output format - json or csv

Optional Parameters

Columns (array, default: ["day", "application"]):

Dimensions: day, hour, application, package_name, ad_format, country, platform, network, network_placement, max_placement, max_ad_unit_id, custom_network_name, ad_unit_waterfall_name, device_type, store_id, has_idfa, max_ad_unit_test
Metrics: impressions, responses, requests, attempts, estimated_revenue, ecpm, fill_rate

Filters:

filter_application: Filter by application name
filter_package_name: Filter by package name
filter_ad_type: Filter by ad type (e.g., banner, inter, rewarded)
filter_country: Filter by two-letter ISO country code
filter_platform: Filter by platform (android, ios)
filter_network: Filter by network name
filter_zone: Filter by zone

Sorting:

sort_day: Sort by day (ASC/DESC)
sort_hour: Sort by hour (ASC/DESC)
sort_estimated_revenue: Sort by estimated revenue (ASC/DESC)

Pagination:

limit: Limit number of results
offset: Skip first N results

Other:

not_zero (boolean): Exclude rows where all metrics are zero

Example Usage

Get revenue data for the last 7 days by country

Claude will use:

{
  "start": "2025-11-25",
  "end": "2025-12-02",
  "format": "json",
  "columns": ["day", "country", "estimated_revenue", "impressions", "ecpm"]
}

2. Cohort Request (`cohort_request`)

Analyzes user cohort performance segmented by installation date.

Required Parameters

start (string): Start date (installation date) in YYYY-MM-DD format
end (string): End date in YYYY-MM-DD format (max 45-day range)
format (string): Output format - json or csv

Optional Parameters

Cohort Type (string, default: revenue):

revenue: Ad revenue and IAP metrics
impression: Ad impression data
session: Retention and session metrics

Columns (array, default: ["day", "installs"]):

Dimensions: day, installs, country, platform, package_name, application
Revenue Metrics: ads_rpi, iap_rpi, pub_revenue, inter_rpi, banner_rpi, reward_rpi
Impression Metrics: imp, imp_per_user, banner_imp, inter_imp, reward_imp
Session Metrics: retention, sessions, session_length, daily_usage

Cohort Interval (string, required for time-based metrics):

Valid values: 0, 1, 2, 3, 4, 5, 6, 7, 10, 14, 18, 21, 24, 27, 30, 45
Specifies days post-install to track (e.g., 7 returns day 0-7 metrics)

Filters:

filter_country: Two-letter ISO country code
filter_package_name: App package name
filter_platform: Platform (android, ios)
filter_application: Application name

Sorting:

sort_day: Sort by installation day (ASC/DESC)
sort_installs: Sort by number of installs (ASC/DESC)

Pagination:

limit: Limit number of results
offset: Skip first N results

Other:

not_zero (boolean): Exclude rows where all metrics are zero

Example Usage

Show me the 7-day revenue per install for users who installed last week

Claude will use:

{
  "cohort_type": "revenue",
  "start": "2025-11-25",
  "end": "2025-11-28",
  "format": "json",
  "columns": ["day", "installs", "ads_rpi"],
  "cohort_interval": "7"
}

This automatically expands to track ads_rpi_0 through ads_rpi_7 for each install cohort.

Project Structure

Applovin-Max-MCP/
├── cmd/
│   └── main.go              # Application entry point
├── internal/
│   ├── capability.go        # Capability abstraction layer
│   └── toolkit.go           # Revenue and cohort tool implementations
├── go.mod                   # Go module definition
├── go.sum                   # Dependency checksums
├── .gitignore              # Git ignore rules
└── README.md               # This file

Development

Building

go build -o bootstrap cmd/main.go

Dependencies

Key dependencies:

github.com/mark3labs/mcp-go v0.43.0 - MCP protocol implementation
github.com/google/uuid - UUID generation
github.com/invopop/jsonschema - JSON schema handling

See go.mod for complete dependency list.

API Endpoints

The server interacts with the following AppLovin Max API endpoints:

Revenue Report: https://r.applovin.com/maxReport
Cohort (Revenue): https://r.applovin.com/maxCohort
Cohort (Impression): https://r.applovin.com/maxCohort/imp
Cohort (Session): https://r.applovin.com/maxCohort/session

Example Queries

Once integrated with Claude, you can ask natural language questions like:

"How much revenue did we make in the last 7 days?"
"Show me the top 5 countries by revenue this month"
"What's our eCPM trend over the past 30 days?"
"Analyze retention rates for users who installed last week"
"Compare iOS vs Android revenue performance"
"Show me rewarded video impressions by application"

Error Handling

The server provides detailed error messages for:

Missing required parameters
Invalid date formats or ranges
API authentication failures
HTTP request failures
Invalid cohort intervals

Security Notes

⚠️ This is a POC with basic security only:

Store your API key securely (use environment variables, never commit to version control)
The API key is transmitted via HTTPS query parameters to AppLovin's secure endpoints
Ensure proper access controls on the bootstrap binary and configuration files
For production use: Implement proper secrets management, request authentication, rate limiting, and audit logging

Production Readiness Checklist

If you want to use this project as a foundation for production, consider implementing:

Version

Current version: 1.0.0 (Proof of Concept)

License

This project is licensed under the MIT License - see the file for details.

Summary: You are free to use, modify, and distribute this software for any purpose, including commercial use, with no warranty provided.

Contributing

Contributions are welcome! This is an open-source POC project. Feel free to:

Fork the repository
Create a feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request

Please ensure any contributions maintain the spirit of simplicity for a POC project.

Support

For issues related to:

This MCP server: Open an issue in this repository
AppLovin Max API: Contact AppLovin support
Model Context Protocol: Visit modelcontextprotocol.io

FurkanEmiroglu/Applovin-Max-MCP-Server