App Store MCP Server for Claude, Cursor & AI Agents

By AppSigma Team · Published July 2, 2026

You can already pull App Store data over REST. But the consumer is increasingly an AI agent: Claude Code auditing your keyword rankings, a Claude Desktop chat triaging last night's 1-star reviews, a Cursor session researching a competitor. For any of them, REST means writing glue: fetch, paginate, paste JSON into the context window, repeat. The Model Context Protocol removes the glue. Connect once, and the model calls the App Store the way it calls any other tool.

So today we're shipping the App Store MCP server: the entire AppSigma App Store API, exposed as MCP tools at a single endpoint.

https://api.appsigma.io/mcp

Connect it to Claude, Cursor, or any MCP client, and you can just ask: "Pull the newest 1-star reviews for YouTube and tell me what people are complaining about." The model fetches the live data itself. No fetch script, no pasted JSON, no glue.

What's in the box

One tool per REST endpoint: 22 in total, covering the full /v1 surface.

  • Apps and metadata: search_applications (real ranked search results, sponsored slots included), get_application, list_applications, get_application_versions, get_application_similars, get_application_in_app_purchases
  • Reviews and ratings: get_application_reviews (filterable by score, country, sort order), get_application_review_histogram, get_application_rating_histogram, get_application_ratings_by_version
  • Charts and prices: get_chart_positions, get_chart_history, get_application_chart_history, get_application_price_changes
  • Developers, events and editorial: get_developer, get_developer_applications, get_developer_events, get_application_events, get_application_editorial_items, get_event, get_editorial_item

The payloads are identical to the REST API's — same fields, same pagination, same identifiers (Apple's numeric track ID). If you've read the App Store API guide, you already know what every tool returns. Nothing new to memorize.

Connect in one minute

If you use Claude Code, it's one command:

claude mcp add --transport http appsigma https://api.appsigma.io/mcp

For Claude Desktop, Cursor, or any other MCP client, add this to the client's MCP config:

{
  "mcpServers": {
    "appsigma": {
      "type": "http",
      "url": "https://api.appsigma.io/mcp"
    }
  }
}

There is no credential in either snippet. On the first call the server tells your client where to sign in, the client opens AppSigma in your browser, and you approve the connection. What it gets back is an ordinary AppSigma API key, so the connection shows up on your API keys page and you can revoke it there like any other.

For a script, a shared machine, or CI, send the key as an X-API-Key header instead and skip the browser step. Same key, same credits, same limits; there's no separate MCP registration, plan, or key type.

Try it without signing up

The demo key skips the sign-in. Add "headers": { "X-API-Key": "demo" } to the config above and you get free, unbilled access to a curated slice of real App Store data: the apps YouTube and Facebook, the search terms youtube, instagram, spotify, and facebook, and the charts (top 10 positions, 3 days of history). It's enough to connect a client and watch the model work before you bother creating a real key.

A good first prompt to test the connection:

Search the App Store for "spotify" and show me the top 5 results with their ratings.

The model calls search_applications, gets the same ranked list a user sees in the App Store, and answers from live data instead of guessing from its training set.

What a session looks like

Once connected, App Store questions become one-liners. Some real prompts and the tools they trigger:

  • "What are users complaining about in the latest version of my app?" → get_application_reviews sorted newest-first, filtered to low scores
  • "Did our rating drop after the last release?" → get_application_ratings_by_version
  • "Who moved into the Top 10 free apps this week?" → get_chart_positions plus get_chart_history
  • "List every app this developer ships and how each is rated" → get_developer_applications
  • "Has this competitor changed their price recently?" → get_application_price_changes

The interesting part is composition — ask "Compare my app's review sentiment against its three closest competitors" and the model chains get_application_similars → get_application_reviews across four apps → synthesis, with no orchestration code on your side.

Billing: same credits, same key

MCP tool calls are billed exactly like the REST requests they wrap. Same credit balance, same rate limits, same cost rules: 1 credit for a single-item lookup, 2 for histograms, 1 per row on lists, 5 for search. Each tool's description states its credit cost, so the model itself can see what a call costs before it makes the call. Demo-key calls are unbilled.

There's no separate MCP pricing because, under the hood, there's no separate MCP implementation: every tool call dispatches through the same authentication, rate-limiting, billing, and validation pipeline as the REST endpoint it wraps. The transport is stateless Streamable HTTP — every request is self-contained, nothing to keep alive, nothing to reconnect.

Where to go next

The App Store MCP page has the short version of everything above. For endpoint-level detail (every field, every parameter, plus a try-it console), go to docs.appsigma.io; the MCP tools mirror those endpoints one-to-one. And if you'd rather call the API directly from code, the REST guide walks the same surface with runnable curl examples.

Grab a key from your AppSigma organization settings, wire up the server, and ask it something real.

Frequently asked questions

Do I need a separate subscription for the MCP server?

No. One AppSigma account covers both surfaces, and a browser-approved connection spends from the same balance as a pasted key. See the pricing section above for what each call costs.

Can I try the App Store MCP server without an API key?

Yes, two ways. Adding the server without any credential sends you through an AppSigma sign-in, which is free and starts with credits. To skip the account entirely, set the header X-API-Key: demo and you reach the sample slice described above, unbilled.

Which clients work with the App Store MCP server?

Any MCP client that supports Streamable HTTP: Claude Code, Claude Desktop, Cursor, ChatGPT, and most agent frameworks. Add the server URL https://api.appsigma.io/mcp, approve the connection when your client opens the browser, and the tools appear automatically. Clients that send custom headers can pass an X-API-Key instead.

Is the MCP server different from the AppSigma REST API?

No, it's the same API over a different transport. Each of the 22 tools wraps one REST endpoint and returns the identical JSON payload, so anything the REST API can do, an AI agent can do through MCP.

Start tracking your keywords →