AppLovin MCP Server (read + write)

An MCP server for AppLovin, covering both sides: Ads Manager (AppLovin's advertiser platform, also called Axon) through the Campaign Management API and Reporting API, and MAX (mediation and monetisation) through the Revenue Reporting, Cohort, User-Level Ad Revenue and Ad Unit Management APIs. Use it from Claude, ChatGPT, Cursor, Codex or any MCP client to report on spend, ROAS and ad revenue, change budgets and goals, manage creative sets and uploads, and tune MAX ad units, floors and waterfalls.

It's maintained by AI Shopping Feeds. It offers:

• both sides of AppLovin in one server: Ads Manager campaigns, creative sets, assets and reports (applovin_ads_*), and MAX revenue, cohort and user-level reports, ad units, waterfalls, experiments and Ad Review rules (applovin_max_*); turn on one side with APPLOVIN_TOOL_GROUPS=ads or =max

• safety: dry-run, confirm words for removals, big batches and budget, goal or CPM jumps, a local audit log with before-snapshots, read-only mode, and guards that keep you clear of AppLovin's 24-hour API block

• an escape hatch for any other AppLovin endpoint, reads open and writes behind confirm="APPLY"

AppLovin publishes its own AppLovin MCP (Node.js, limited beta, Ads Manager campaigns and reporting). This one is Python, adds MAX, and puts dry-run, confirm words and an audit log in front of every write; see the comparison. Not affiliated with or endorsed by AppLovin.

Part of the AI Shopping Feeds MCP family: Google Ads · Google Merchant Centre · Google Analytics · YouTube Studio · Amazon Ads · Apple Ads · Microsoft Ads · TikTok Ads · Pinterest Ads · Snapchat Ads · OpenAI Ads · AppLovin.

Let your agent set it up

Paste this into Claude Code, Codex, Cursor or any agent that can read web pages and edit files:

Set up the AppLovin MCP server for me by following https://raw.githubusercontent.com/AI-Shopping-Feeds/applovin-mcp/main/skills/applovin-mcp/SKILL.md

The agent checks whether it's already connected, installs the runner if you agree, tells you which AppLovin keys to copy for the side you use, has you store them with applovin-mcp setup (so they never pass through chat), adds the server to your client's config without touching your other servers, and checks it works by calling applovin_health_check.

The same file is an Agent Skill, so your agent can keep it as a guide to using the tools well. For Claude Code, copy the folder into your skills:

git clone --depth 1 https://github.com/AI-Shopping-Feeds/applovin-mcp.git /tmp/applovin-mcp
cp -r /tmp/applovin-mcp/skills/applovin-mcp ~/.claude/skills/

Other agents that support skills load the same skills/applovin-mcp folder from their own skills directory.

Related MCP server: google-ads-mcp

What it can do

42 tools, all prefixed applovin_. The full list is in docs/tools.md.

Read

Write

Safety

• Dry-run (APPLOVIN_DRY_RUN=true or applovin_set_dry_run) records writes and shows what would be sent, without sending.

• Confirm words. Removals need confirm="DELETE". More than 10 items, daily budget, CPI/CPE/CPP goal, floor or CPM rises above 20%, ROAS goal cuts above 20%, values whose current figure can't be read, disabled networks, experiment promotions and raw writes need confirm="APPLY". Optional caps: APPLOVIN_MAX_DAILY_BUDGET, APPLOVIN_MAX_CPM.

• Audit log. Every write goes to ~/.applovin-mcp/logs/audit.jsonl (0600) with a before-snapshot, and is pruned after APPLOVIN_AUDIT_RETENTION_DAYS (30). applovin-mcp purge deletes it and the stored keys. Report rows stay in memory.

• Read-only mode (APPLOVIN_READ_ONLY=true, or AppLovin's APPLOVIN_READONLY) and tool groups (APPLOVIN_TOOL_GROUPS=ads, max or group names).

• AppLovin's limits. The Campaign Management API allows 1000 requests a minute and blocks an account for 24 hours after 100 failed requests in five minutes; the server checks arguments first, stops batches at the first failure and pauses those calls if errors pile up.

• Privacy. User-level revenue is summarised in memory with IDFA, IDFV and user IDs masked unless you ask for them. Uploads only read from APPLOVIN_UPLOAD_DIR.

• The HTTP transports refuse a non-loopback address unless APPLOVIN_ALLOW_REMOTE=true.

• What AppLovin's terms allow, and how this server follows them: docs/terms-and-data.md.

Supported clients

Setup

1. Install uv

The server runs with uvx from uv (brew install uv on macOS) and needs Python 3.11 or later. It installs straight from this repository:

uvx --from git+https://github.com/AI-Shopping-Feeds/applovin-mcp.git applovin-mcp

Pin a release by adding @vX.Y.Z to the URL.

2. Get credentials

AppLovin uses long-lived keys, one per API. You only need the side you use. Step by step, including who can see the keys and how to get whitelisted: docs/setup-applovin-api.md.

1. Ads Manager (advertisers): an Admin opens Ads Manager → account menu → Keys and copies the Report Key (reports) and, if AppLovin has whitelisted the account, the Campaign Management API key. The account ID is under account menu → Go to settings.

2. MAX (publishers): in the MAX dashboard, Account → General → Keys: the Report Key (reports), the Management Key (ad units) and, for Ad Review rules, the Ad Review Key.

3. Store them without pasting them into chat:

   uvx --from git+https://github.com/AI-Shopping-Feeds/applovin-mcp.git applovin-mcp setup

   This asks for each key without echoing it and saves them to ~/.applovin-mcp/secrets/keys.json (owner-only). You can use environment variables instead; they win over the file.

3. Add it to your client

Replace the placeholders. Keep your other servers in each file. With keys stored by setup, only the options you want need setting; this example limits the server to MAX.

Claude Code

claude mcp add --scope user applovin \
  --env APPLOVIN_TOOL_GROUPS=max \
  --env APPLOVIN_DRY_RUN=true \
  -- uvx --from git+https://github.com/AI-Shopping-Feeds/applovin-mcp.git applovin-mcp

Claude Desktop, Cursor, Gemini CLI, Windsurf (and VS Code under servers)

{
  "mcpServers": {
    "applovin": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/AI-Shopping-Feeds/applovin-mcp.git", "applovin-mcp"],
      "env": {
        "APPLOVIN_ACCOUNT_ID": "123456789",
        "APPLOVIN_UPLOAD_DIR": "/abs/path/creatives",
        "APPLOVIN_DRY_RUN": "true"
      }
    }
  }
}

Desktop apps often can't find uvx on the PATH; if the server doesn't start, use the full path from which uvx as the command.

Codex (~/.codex/config.toml)

[mcp_servers.applovin]
command = "uvx"
args = ["--from", "git+https://github.com/AI-Shopping-Feeds/applovin-mcp.git", "applovin-mcp"]
env = { APPLOVIN_ACCOUNT_ID = "123456789", APPLOVIN_DRY_RUN = "true" }

4. Check it works

Restart or reload the client, then ask it to "check my AppLovin connection". It should call applovin_health_check and show one line per key: a 401/403 on the Campaign Management API means AppLovin hasn't whitelisted the account yet, while reports still work. Turn dry-run off when you're ready to make changes.

Configuration

Settings are read from the environment, then ./.env (or --env FILE), then ~/.applovin-mcp/.env. A copy of every variable is in .env.example.

Example prompts

Show last week's spend, installs and 7-day ROAS by campaign in Ads Manager, worst first.

Raise the daily budget of every campaign with 7-day ROAS above target by 10%. Rehearse it in dry-run first and show me the changes.

Which MAX networks earned the most on my rewarded ad units yesterday, and how did fill rate compare by country?

Compare day-7 retention and revenue per install for installs from the last two weeks, by country.

More docs

• All tools

• Getting AppLovin API keys

• Terms, data handling and the official AppLovin MCP

• Agent skill: setup and usage guide for agents

• Changelog and security policy

Contributing and licence

Issues and pull requests are welcome on this repository. See CONTRIBUTING.md.

Licensed under the MIT License.

Try out AI Shopping Feeds, our shopping feed management tool (including Google Shopping feeds).