Bring AppDrift into Claude, Cursor, and any AI agent. Keyword difficulty, live ranks, page-one opportunities, and your weekly action plan — read directly by the agent you already build in. The first ASO platform your agent can drive.
Capabilities
Five read-only tools over one API — through MCP or plain REST.
Score any App Store or Google Play keyword — difficulty and popularity — straight from your agent. Works for any keyword, not just the ones you track.
Read your apps' tracked keywords and their latest store rank. Your agent can answer "where do I rank for meditation?" without you opening a dashboard.
Surface the keywords ranked 11-30 — the ones one good push away from page one — ordered easiest-to-move first. The highest-ROI work, ready to act on.
Your whole prioritized ASO plan — keyword pushes, rank drops, review themes, competitor overtakes, localization gaps — as structured data an agent can reason over.
A first-class Model Context Protocol server. Add it once and the AppDrift tools appear inside Claude Desktop, Cursor, and any MCP client you already use.
Prefer plain HTTP? Every capability is a simple GET with a bearer key — call it from any language, script, CI job, or automation platform.
Two ways in
Connect once and ask away — or curl it from anywhere. Full setup and the complete endpoint reference are in the docs.
{
"mcpServers": {
"appdrift": {
"command": "npx",
"args": ["-y", "appdrift-mcp"],
"env": { "APPDRIFT_API_KEY": "ad_live_..." }
}
}
}curl -H "Authorization: Bearer ad_live_..." \
"https://appdrift-backend-1fabfc95f592.herokuapp.com/v1/action-plan"How it works
You'll be asking your agent about your ranks before your coffee cools.
Open the Developers area of your dashboard and mint a key. It's shown in full once, tenant-scoped, and revocable any time.
Drop a few lines of JSON into your MCP client config, or point any HTTP client at the REST API with your key as a bearer token.
"What should I optimize this week?" Your agent reads your real ranks, opportunities, and action plan — and answers.
FAQ
It's a Model Context Protocol server that lets Claude, Cursor, and any MCP-compatible agent read your ASO data directly — keyword difficulty, tracked ranks, page-one opportunities, and your weekly action plan. Instead of switching to a dashboard, you ask your agent "what should I optimize this week?" and it answers from your real data.
Create an API key in the Developers area of your dashboard, add the AppDrift MCP server to your client config (a few lines of JSON), and restart the client. The full copy-paste setup lives in the MCP & API docs. You can also call the read-only REST API directly with any HTTP client.
Yes. The v1 API and MCP server surface your data but never change your store listings — write actions stay in the AppDrift dashboard, so an agent can read and reason but can't accidentally publish. Charging-based generation tools arrive in a later version.
It's included on every paid AppDrift plan, starting with a 7-day free trial. Keys are tenant-scoped and revocable at any time from the Developers page — a key only ever sees the account it was created in.
Anything that speaks the Model Context Protocol — Claude Desktop, Cursor, and the growing ecosystem of MCP clients. For everything else, the plain REST API works with any language, script, or automation platform.
Grab a key and wire AppDrift into Claude or Cursor in a couple of minutes.
One-shot AI text generation is commoditizing fast — any founder can ask a chatbot for an app description. What can't be answered from a blank chat window is your live, longitudinal ASO reality: where you actually rank today, which keywords are one push from page one, what your reviewers keep complaining about, and which competitor just overtook you. That data lives in AppDrift. The MCP server and REST API put it inside the agent you already build with, so optimization happens in the flow of work instead of in yet another dashboard tab.
The Model Context Protocol is the emerging standard for giving AI agents safe, structured access to external tools and data. AppDrift ships a first-class MCP server: add a few lines to your Claude Desktop or Cursor config, drop in an API key, and the AppDrift tools appear alongside everything else your agent can do. Ask "which of my keywords are closest to page one?" and it calls keyword tracking for you. Ask "what should I ship this week?" and it reads your ASO Autopilot action plan and summarizes the top moves.
The v1 API is deliberately read-only. It surfaces your data — keyword difficulty, ranks, opportunities, the weekly plan — but never touches your live App Store or Google Play listings. An agent can read everything and reason freely, but it can't accidentally publish a half-baked description to production. Write actions stay in the AppDrift dashboard, behind human review, exactly where they belong. It's the safe way to give an autonomous agent real leverage over your ASO without handing it the keys to your store presence.
The surface is small on purpose. Keyword difficulty scores any keyword on demand. Rank tracking returns your tracked keywords and their latest positions. Opportunities surfaces the keywords ranked 11-30 that are cheapest to move onto page one. The action plan hands over your whole prioritized to-do list — pushes, rank drops, review themes, competitor overtakes, localization gaps — as structured data. And list apps ties it together so the agent knows which apps it's reasoning about. Authentication is a single tenant-scoped key you can revoke any time; a key only ever sees the account that created it.
Not every workflow runs through an agent. Every MCP tool is backed by a plain REST endpoint — a simple authenticated GET — so you can pull keyword difficulty into a spreadsheet, wire rank checks into a CI job, post opportunities to Slack, or feed the action plan into your own BI stack. Same data, same key, any language. When you're comparing AppDrift to other ASO platforms, this openness is part of the story: see how it fits the wider toolkit in our rundown of ASO tools in 2026, or profile any app on the stores with App Intelligence. API and MCP access is included on every paid plan — see pricing to get a key.