ASO MCP server

An ASO MCP server
on your own data.

Storewatch’s MCP server gives an AI agent five ASO tools: the 0–100 audit, Markets, keyword ideas, tracked keywords with rank, popularity and difficulty, and your listing text. It reads your own App Store numbers on your Mac and can draft changes. It never publishes.

3 days freeOne-time purchaseLoved by developers

Markets for the sample Calorie Tracker app over 90 days, with Ask Storewatch answering which country to localize next: Japan, the biggest market still shown the English listing, with Mexico as runner-up, a warning not to expect a big conversion jump, the US listing’s audit score of 40, and an offer to draft the Japanese listing for review.

What “ASO MCP server” means here.

Most ASO MCP servers sit in the cloud and answer from a market database. Storewatch’s sits on your Mac and answers from your own apps: the keywords you track, the ranks checked in each country, your impressions and downloads, and your listing text. An AI agent such as Claude Code, Codex or Cursor asks it questions and gets compact JSON back.

Five of its tools are about ASO. This page names them, shows what each returns and what an agent can do with the result. For turning the server on, the three access levels and the other tools, see the App Store Connect MCP page.

The five ASO tools, and what each returns.

Every tool takes an app_id from list_apps. The data comes from Storewatch’s local database, so an answer arrives in seconds and covers weeks of history.

ToolArgumentsWhat the agent gets back
get_aso_auditapp_id, optional countryThe overall score, then per country a 0–100 score with its three parts (rankings out of 50, listing out of 30, competition out of 20), the listing language and whether it was written for that country. Each tracked keyword comes with rank, difficulty, popularity and where it sits: “title or subtitle”, “keyword field” or “missing”. Last, the fixes worth doing first, each with a title and a detail.
get_marketsapp_idUp to 40 countries by impressions and downloads over the last 90 days: whether each has a listing in its own language, which language, keywords tracked, the audit score and a next step such as “write a German listing”, “track keywords”, “improve the listing” or “nothing urgent”.
get_keyword_ideasapp_id, country (required)Up to 60 phrases you don’t track yet. Each has a reason, plus popularity, difficulty and your rank when Storewatch already has them. They come from the app’s own listing and description, and from words used by several of the top apps in your tracked searches.
get_keywordsapp_id, optional countryUp to 150 tracked keywords with country, rank, difficulty and popularity. A popularity_source field says whether the number is from Apple Ads (0 means too few searches to list), imported, or unknown.
get_listingapp_id, optional localeTitle, subtitle, keywords, promotional text, description and What’s New for one language, the languages available, the version being prepared and any drafts you have not published.

Ranks are the last ones Storewatch checked in the App Store for each country. A keyword added a minute ago has no rank until the next sync.

Ask what to fix.
Get real answers.

The agent chains the tools itself: markets, then a country’s audit and keyword ideas, then the answer.

The Keywords screen filtered to Germany, with kalorienzähler at #9 and its rank history, and Ask Storewatch answering which keywords to add for Germany: German words already in the keyword field but not tracked, words the top competing apps use, a note that the ideas have no popularity data until they’re tracked, and the 39 of 100 keyword characters in use.

Questions it can answer.

These need your numbers, not a market database. Each one maps to tools you can see in the table above.

AskTools it uses
Which country should I localize next?get_markets, then get_aso_audit for the top pick
Which of my keywords are missing from the title, subtitle and keyword field?get_aso_audit, reading each keyword’s “missing” flag
Which tracked keywords rank 11 to 30 and are easy to move?get_keywords, sorted by rank and difficulty
What should I track in France that I don’t yet?get_keyword_ideas for fr, then add_keywords if you agree
Did last month’s subtitle change bring more impressions?get_events, compare_before_after, get_totals
Is my German listing written for Germany or only translated?get_aso_audit, reading listing_written_for_country

An agent that reads the audit first has the fixes in front of it, so it doesn’t guess. The audit tool’s own description tells the model to use it before suggesting title, subtitle or keyword changes.

Drafts only.
Nothing sent.

With Full access the agent drafts listing text within Apple’s limits. You publish it, or you don’t.

The Listing editor with Changes to publish: the title, subtitle, keywords and promotional text, each showing the old and the new text, with character counts, Discard All, Publish and History below.

What it can change, and what it never does.

Action toolWhat it doesWhere it stops
draft_listing_changesWrites title, subtitle, keywords, promotional text, description or What’s New, in any language the app hasLimits enforced: 30, 30, 100, 170, 4,000 and 4,000 characters. Saved as a draft. The agent has no publish tool
add_keywords, remove_keywordsStarts or stops tracking keywords in the countries you nameChanges tracking in Storewatch only. A new rank appears on the next sync
log_eventMarks a listing change, price change, promotion, ad campaign or note on your chartsLocal to Storewatch. Used later for before-and-after checks

An agent can’t publish a listing, submit a version for review, change a price in App Store Connect or see your App Store Connect key. It only reaches what the local helper exposes. With Read only access it can’t draft at all, and with Off it answers nothing.

How it compares with hosted ASO MCP servers.

Storewatch is made by us, so read this as one side. Hosted servers have something a local one doesn’t: market data for apps that are not yours. Storewatch has something they don’t: your own App Store Connect numbers next to your ranks. Details as of November 2026, from each company’s own page.

StorewatchSonarWow KeywordsAppTweak
Where it runsLocal helper on your Mac, reads your synced dataHosted MCP server, 80 toolsDesktop app, MCP with 116 tools; keyword database stored on your computerHosted, with an API and MCP
Keyword dataRanks for the keywords you track, Apple Ads popularity, difficulty from the search results, ideas from your listing and top appsDaily ranks across 72 marketsRank, volume and difficulty in 52 countries; volume estimated from ratings of matching appsKeyword research, popularity, difficulty and daily ranks
Your own sales and impressionsYes, from App Store ConnectNo sales or analytics listedNone found on its pagesNone found on its page
Competitor trackingIn the app, not in the MCP toolsNot listed on its pageCompetitor ranks side by sideCompetitor tracking
Price$29 once ($49 from December 2026)From $24.17 a month billed yearly; $29 monthly$99 once, rising as licenses sellFrom $79 a month

Pick a hosted server if you need to look at other people’s apps at scale. Pick Storewatch’s if you want an agent that works on your apps with your real downloads. They don’t exclude each other: you can connect both to the same AI app.

Questions, answered.

Do I need Apple Ads for the popularity numbers?

Popularity comes from Apple Ads once you connect an Apple Ads key, which a free account can provide. Without it, get_keywords marks the number as imported or unknown through its popularity_source field.

Can the agent rewrite my listing in another language?

Yes, as a draft. It reads the listing with get_listing, writes the new text for a language the app has and calls draft_listing_changes. It must fit each field’s limit, and you publish it from Storewatch.

Does it show competitors’ keyword ranks?

Not through MCP. The get_keyword_ideas tool uses words from the top apps in your own tracked searches. Rank and listing tracking for named competitors is in the app’s ASO → Competitors screen.

Which AI apps can use these tools?

Any app that can run a local MCP server. Settings → AI has one-click setups for the Claude app, Claude Code, Codex and Cursor, and a JSON config for others.

Is my data sent anywhere?

The helper reads your Mac. The AI app then sends what a tool returns to its own provider so it can answer. If that matters, use a model that runs on your Mac.

Sources: Storewatch’s MCP server and Settings → AI, version 3.5 (November 2026). Apple’s limits: App Store Connect Help. Competitors, checked November 2026: Sonar, Wow Keywords, AppTweak. MCP is described at modelcontextprotocol.io.

Related: App Store Connect MCP server · AI agent · ASO audit · Keyword research · Claude Code and the App Store · Storewatch vs Wow Keywords

Made by Alex, the developer of Storewatch. Last updated November 21, 2026.

Give your AI your ASO data.

3 days freemacOS 15 or later on Apple silicon