Best ASO MCP Servers Compared (2026)

Which ASO MCP Server Should You Use in 2026?

The Model Context Protocol — open-sourced by Anthropic in late 2024 and donated to the Linux Foundation's Agentic AI Foundation in December 2025 (source: Wikipedia, Model Context Protocol) — now has over 10,000 active servers in production (source: MCP Adoption Statistics 2026, Digital Applied). A handful of those are built specifically for App Store Optimization. They let Claude, Cursor, and other AI agents research keywords, audit metadata, and pull review sentiment without leaving your editor.

The category is young and fragmented. Some of these servers return scraped storefront data; others return computed difficulty scores from proprietary indexes. Some run only locally; several are hosted. I tested every server in this category I could find in July 2026 and compared them on data quality, store coverage, hosting model, free tier, write tools, and tool count.

The category also changed in July 2026: four established ASO platforms — Appfigures, AppTweak, Sensor Tower and AppFollow — shipped official MCP servers within weeks of each other, turning this from an indie-tooling niche into something the incumbents now ship by default.

Below are the ten worth knowing about — starting with Sonar (trysonar.app; disclosure: I built it), then the official vendor servers, and ending with early-stage prototypes that cover gaps the others miss.

What Is an ASO MCP Server?

An ASO MCP server exposes App Store Optimization data — keyword difficulty, search popularity, competitor metadata, review sentiment — as callable tools for AI agents. Instead of switching between a browser-based ASO dashboard and your IDE, you ask Claude or Cursor "What's the difficulty for 'budget planner' on iOS?" and the server returns structured data inline.

The key differentiator between servers is data source. Scrapers pull raw storefront HTML and return whatever Apple or Google show on the public page. Computed-metric servers maintain their own keyword indexes and return proprietary scores like difficulty and popularity that don't exist in the raw store data. This distinction determines whether you get "here are the top 10 results for this keyword" or "this keyword has difficulty 66 and Apple-reported popularity 51 with 191 competing results" — the kind of competitive intel an ASO keyword research workflow actually needs.

The Comparison Table

1. Sonar (@sonarapp/mcp) — 47 Tools, Computed Metrics, Both Stores, Hosted

Sonar's MCP server returns computed keyword metrics — difficulty scores, Apple-reported popularity, and result counts — rather than raw scraped data. The official AppTweak and Sensor Tower servers also return computed platform metrics, but both require a paid API or enterprise subscription; Sonar is the one you can point an agent at on a free tier. It ships as an npm package (@sonarapp/mcp, source at github.com/trysonar/mcp) and connects to Sonar's hosted REST API, so there is no local scraping and no rate-limit risk from Apple or Google (source: glama.ai/mcp/servers/trysonar/mcp). It is listed in the official MCP registry, Smithery, and Glama.

Install:

npx -y @sonarapp/mcp@latest

Or in an MCP client config:

{ "command": "npx", "args": ["-y", "@sonarapp/mcp@latest"] }

No API key is required to start — with SONAR_API_KEY unset the server runs in keyless free mode, exposing the stateless research tools against the anonymous API tier (30 requests/day per IP, plus difficulty and popularity for 5 keywords/day). Set a key to lift the limits and unlock workspace and write tools.

Where the iOS popularity number comes from: Sonar serves the real Apple search popularity score — the same 5–100 scale Apple shows in its own Search Ads console — for iOS keywords, and you do not need your own Apple Search Ads account to read it. A calibrated estimator fills in keywords below Apple's coverage and all Android keywords.

What it returns that scrapers cannot: Sonar's keyword index puts "tip calculator" at iOS difficulty 39 with Apple popularity 5, vs Android difficulty 16 — illustrating the cross-platform gap Sonar's MCP server surfaces in a single query (source: Sonar /api/v1/keywords/search, queried 2026-07-23). For "budget planner," Sonar reports iOS difficulty 66, Apple popularity 51, and 191 competing results — the kind of competitive intel the server returns in under 2 seconds (source: Sonar /api/v1/keywords/search, queried 2026-07-23).

If you're unfamiliar with what those difficulty numbers mean in practice, our guide to keyword difficulty scores breaks down the methodology.

Tools (47 total): the highest count in this roundup, and full parity with Sonar's v1 REST API. Stateless research tools (keyword search, app lookup, ASO audit, keyword extraction, review mining, revenue estimation, top charts) work keyless or on credits across both iOS and Android. Workspace reads — 90-day app snapshots, 365-day keyword ranking history, competitor change detection — require a subscription. Write tools (product creation, keyword tracking up to 200 keywords, screenshot management) require a subscription plus a write-scope API key (source: trysonar.app/connect-ai/cursor).

Pricing: free keyless mode, plus 50 free API credits on signup. Prepaid credit packs start at $10 for 1,000 credits with no subscription required. The Indie plan ($19/month or $189/year) unlocks workspace features, with a 7-day free trial (source: trysonar.app/agents).

Best for: Teams that need cross-platform keyword data in agentic workflows — difficulty scores, popularity, and competitor intel — without running a local scraper. The hosted model means the same MCP config works in Claude Desktop, Claude Code, Cursor, or any MCP client with zero infrastructure.

Explore Sonar's MCP server →

2. The Official Vendor Servers — Appfigures, AppTweak, Sensor Tower, AppFollow

Between mid-July and late July 2026, four established ASO platforms shipped official MCP servers. If you already pay for one of these tools, its server is almost certainly the fastest path to agent access — you are not adding a vendor, just switching interface. The catch is uniform: none of them has a free tier, because each rides on a paid API or platform subscription.

Appfigures publishes the only vendor server that ships alongside a real CLI. Both live in the same package — npm install -g @appfigures/cli — and expose app metrics, reviews, store data and organic keyword rankings "with position, popularity, and competitiveness" (source: github.com/appfigures/cli). Note the metering: CLI and MCP calls cost 10 API credits each, versus 5 for a regular ASO request (source: Appfigures Knowledge Base). Agent workflows are chatty, so that doubling adds up fast.

AppTweak runs a remote server at https://developers.apptweak.com/mcp offering "Direct API access to AppTweak API functionality" and "Real-time data from your AppTweak API account" (source: developers.apptweak.com/reference/mcp). It also answers documentation questions, which is genuinely useful when you are writing integration code. AppTweak sells its REST API as a separate product from the platform tiers, so budget for both.

Sensor Tower announced its server in July 2026: "Now, you can easily access Sensor Tower and Pathmatics data directly from your company's chatbot tool." It "validates the request, and translates it to the Sensor Tower API backend" — real data, not documentation — and is "an additional product for qualifying Sensor Tower users" (source: sensortower.com/blog/introducing-the-sensor-tower-mcp-server). This is the only one on the list where you cannot find out the price without a sales call.

AppFollow ships an AI toolkit around a live server at https://mcp.appfollow.io/mcp, described as turning "the live AppFollow MCP server into ready-made analyst workflows — for Claude Code and Codex" (source: github.com/AppFollow/appfollow-ai-toolkit). Given AppFollow's strength is review management, this is the one to reach for if your agent workflow is about review sentiment rather than keyword research.

Best for: Existing customers of any of these platforms. Watch for: per-call metering (Appfigures), a separately-priced API (AppTweak), and no published pricing at all (Sensor Tower).

3. KenanAtmaca/aso-mcp — Open-Source, iOS-Focused, App Store Connect Integration

KenanAtmaca's aso-mcp is a TypeScript MCP server that pulls iOS App Store data using the app-store-scraper and aso npm packages. It requires no API key for read operations and runs locally via stdio transport (source: github.com/KenanAtmaca/aso-mcp).

Data quality: The server scrapes live App Store results and computes scores using the aso npm package. When Apple returns 503 errors (which happens frequently enough that the README documents it), the server falls back to custom scoring based on search-result analysis. This means scores can vary between sessions depending on Apple's availability.

Tools (19 total): Keyword search and suggestions, app details, competitor analysis, metadata optimization with character validation, review sentiment analysis, ranking tracking with saved snapshots, localized keyword performance across 155+ countries, and ASO report generation. Phase 5 adds App Store Connect integration — read and update app metadata directly, with a 40-locale batch-update limit.

Limitations: iOS only. No Android coverage. Requires Node.js 22+. App Store Connect features need issuer ID, key ID, and private key credentials. The local-only model means each developer needs their own install and faces Apple rate limits directly.

Best for: Solo iOS developers who want an open-source, no-cost MCP integration for ASO and are comfortable with scraper-quality data. The App Store Connect write integration is a standout feature for developers managing metadata from their editor.

4. appreply-co/mcp-appstore — Lightweight Scraper, Both Stores

AppReply's mcp-appstore is the simplest server in this comparison. It provides six tools for searching apps, pulling details, analyzing keywords, mining reviews, and finding similar apps across both iOS and Android (source: github.com/appreply-co/mcp-appstore).

Data quality: Pure scraper. The server queries Apple and Google store APIs directly and returns raw results — app names, ratings, review text, developer info. It does not compute keyword difficulty, popularity, or any derived metric. The analyze_top_keywords tool returns brand-presence analysis and competition metrics based on current search results, not a historical index.

Tools (6 total): search_app, get_app_details, analyze_top_keywords, analyze_reviews, fetch_reviews (Android-only developer responses), get_similar_apps. Results are cached via memoization for 10 minutes.

Limitations: No difficulty scores. No popularity data. No write tools. No workspace or tracking. Large review pulls can be slow. The server is a pure read-only research tool.

Best for: Quick competitive research where you need to pull app metadata and reviews into an AI conversation without computed metrics. If you need an ASO tool that goes deeper, this server is a starting point, not a destination.

5. AppSprint MCP — 38 Tools, macOS Only, Apple Ads Integration

AppSprint offers the second-highest tool count in this roundup (38 tools, behind Sonar's 47), but comes with a hard constraint: it runs as a local server inside a macOS desktop app. You must keep AppSprint open on your Mac while the MCP server is active (source: appsprint.app/docs/mcp).

Data quality: AppSprint uses its own proprietary keyword data and integrates directly with App Store Connect and Apple Ads. The data is tied to apps you actively track in the AppSprint desktop app, so the MCP server reads from your local database rather than scraping on demand.

Tools (38 total): Across five categories — ASO and App Store Connect (15 tools for keyword management, metadata, screenshot uploads), analytics reads (4 tools for installs, sources, geography, retention), Apple Ads reads (5 tools for campaign and keyword performance), Apple Ads planning (8 tools for bid suggestions, playbooks, diagnostics), and Apple Ads writes (6 tools for campaign creation, bid updates, budget changes, keyword additions).

Write safeguards: All write operations default to dry-run mode and return a diff before executing. Budget increases are blocked unless explicitly approved. This three-layer workflow (read, plan, write) is the most cautious approach of any MCP integration in this comparison.

Limitations: macOS only — no Windows, Linux, or remote server support. iOS only — no Google Play coverage. Pricing is not publicly documented on the MCP docs page. The local-only architecture means you cannot share an MCP connection across a team.

Best for: Solo iOS developers on Mac who already use AppSprint and want to add Apple Ads management to their AI workflow. The Apple Ads integration is unique — no other server in this category offers campaign creation and bid optimization tools.

6. RankoR/aso-mcp — Google Play Only, Python, Prototype

RankoR's aso-mcp is the only MCP server in this roundup built exclusively for Google Play. It is written in Python 3.14+ and uses the play-store-scraper-ng library for data (source: github.com/RankoR/aso-mcp).

Data quality: Pure Google Play scraper. Returns app details, reviews, collection listings (top free, top paid, top grossing), and autocomplete keyword suggestions. The metadata validation tools enforce Google Play character limits (30-character title, 80-character short description, 4,000-character full description) — a useful guardrail when an AI agent is drafting Play Store listings.

Tools (11 total): App search with price filtering, app details, reviews with sorting and pagination, collection and category browsing, keyword suggestions via autocomplete, three metadata validation tools, and reference tools for supported languages and countries across 60+ regions.

Limitations: The README explicitly states: "This MCP is a very early feature-limited prototype and is not yet ready for production use." No iOS support. No difficulty or popularity scores. No write tools. Optional proxy rotation for rate-limit mitigation suggests the scraper can hit Google's throttling.

Best for: Android-focused developers who want a free, open-source tool for Play Store research inside an AI agent and are comfortable with prototype-stage software.

7. astro-mcp-server — Astro ASO Data for LLMs

The astro-mcp-server lets you query Astro's ASO keyword and ranking data from Claude, Cursor, or any MCP client instead of switching to the Astro Mac app. It is a community-built bridge by Tim Broddin — not an official Astro product — that reads the SQLite database the Astro app maintains locally on your Mac (source: github.com/TimBroddin/astro-mcp-server).

Astro ASO MCP Data and Tools

The server exposes four read-only tools that surface the ASO data Astro has already collected for your tracked apps and keywords:

• search_rankings — keyword rankings filtered by app store and app, returning current position data.
• get_historical_rankings — ranking changes over time for a given keyword, useful for spotting trends after a metadata update.
• list_apps — all applications you track in Astro.
• get_app_keywords — keywords with current rank, previous rank, difficulty, and Apple Search Ads popularity scores.

Because these tools read Astro's local database, the data is only as fresh as Astro's last sync. If Astro has not refreshed rankings recently, the MCP server returns stale results. There is no on-demand fetch from the App Store.

Setup and requirements

The server runs locally via stdio transport — add its path to your Claude Desktop or Cursor MCP config and point it at the SQLite database that Astro maintains locally on your Mac. No API key is needed, but Astro must be installed on the same Mac with tracking data already collected, which in practice means an active Astro subscription at $9/month (source: tryastro.app). For a step-by-step walkthrough of connecting ASO MCP servers to Claude, see our Claude MCP setup guide.

Limitations: iOS only — no Google Play coverage. Only 4 tools. No keyword discovery or research — the server surfaces data for keywords and apps you already track in Astro, so it cannot help you find new opportunities. No write tools, so you cannot update metadata from your editor. macOS only. If you're evaluating whether Astro fits your ASO workflow or need cross-platform coverage, see our Astro alternatives comparison.

Best for: Existing Astro subscribers who want to pipe their tracked iOS ranking and keyword difficulty data into Claude or Cursor without opening the Astro Mac app — especially useful for generating periodic ranking reports or analyzing keyword trends conversationally.

How to Choose the Right ASO MCP Server

Picking the right server depends on three factors: what store you target, whether you need computed metrics or raw data, and how much infrastructure you want to manage.

If you need both iOS and Android keyword data with difficulty scores, Sonar is the only option that computes cross-platform metrics from a maintained index rather than scraping. The hosted model eliminates rate-limit concerns. See Sonar's MCP setup docs →

If you're iOS-only and want free, open-source tooling, KenanAtmaca/aso-mcp offers the most complete feature set with App Store Connect integration. Accept that scraper data can be inconsistent when Apple throttles requests.

If you're Android-only, RankoR/aso-mcp is the dedicated option, though its prototype status means you should expect rough edges.

If you manage Apple Ads alongside ASO, AppSprint's 38-tool server is the only one with campaign management — but it locks you into macOS.

If you already pay for Astro, the astro-mcp-server lets you access your existing data through MCP without changing your workflow.

For teams running automated marketing workflows with AI agents, the hosted model matters most. A local scraper that breaks when Apple rate-limits is a pipeline blocker. A hosted API that returns consistent data is infrastructure you can depend on.

> Key takeaway: If you need computed keyword metrics (difficulty, popularity) across both iOS and Android with zero local infrastructure, Sonar is the only hosted option. If you want free, open-source tooling and can tolerate scraper inconsistencies, KenanAtmaca/aso-mcp (iOS) and RankoR/aso-mcp (Android) cover each store individually. AppSprint stands alone for Apple Ads campaign management but locks you into macOS.

Frequently Asked Questions

What is an ASO MCP server?

An ASO MCP server exposes App Store Optimization data — keyword metrics, app metadata, review sentiment — as tools callable by AI agents like Claude, Cursor, or Cline. You query the server from your editor instead of copy-pasting from a dashboard. Anthropic open-sourced MCP in late 2024, and by 2026 the protocol has over 10,000 active servers in production (source: Digital Applied, MCP Adoption Statistics 2026).

Do ASO MCP servers return keyword difficulty scores?

Not all of them. Scraper-based servers like appreply-co/mcp-appstore and RankoR/aso-mcp return raw store data — search results, reviews, app metadata — without computing difficulty or popularity. Sonar's MCP server returns computed difficulty scores and Apple-reported popularity from a maintained keyword index. For example, "tip calculator" shows iOS difficulty 39 and Apple popularity 5 in Sonar's data (source: Sonar /api/v1/keywords/search, queried 2026-07-23). The official AppTweak and Sensor Tower servers also return their platforms' computed metrics, but only to paying API or enterprise customers. AppSprint surfaces its own metrics but only for apps you actively track in its macOS app.

Can I use an ASO MCP server without an API key?

Yes, for some servers. KenanAtmaca/aso-mcp, appreply-co/mcp-appstore, and RankoR/aso-mcp are open-source and require no API key — they scrape store data directly. Sonar's server also runs keyless: with SONAR_API_KEY unset, npx -y @sonarapp/mcp@latest works in free mode against the anonymous API tier (six endpoints, 30 requests/day per IP, plus difficulty and popularity for 5 keywords/day). Adding a key gets you 50 free credits on signup, with prepaid packs from $10 (source: trysonar.app/agents). AppSprint requires its macOS app running locally. The astro-mcp-server needs no key at all — it reads Astro's local database, so it needs the Astro app installed with data already collected. The four official vendor servers (Appfigures, AppTweak, Sensor Tower, AppFollow) all require a paid account.

Which ASO MCP server covers both iOS and Android?

Six do: Sonar (@sonarapp/mcp), appreply-co/mcp-appstore, and the four official vendor servers from Appfigures, AppTweak, Sensor Tower and AppFollow. Sonar provides computed metrics (difficulty, popularity) for both platforms from a single query and is the only one of the six usable on a free tier; appreply-co/mcp-appstore scrapes both stores but returns raw data without computed scores; the vendor servers each require a paid account. The remaining servers are single-platform: KenanAtmaca/aso-mcp and AppSprint are iOS-only, RankoR/aso-mcp is Android-only, and astro-mcp-server is iOS-only.

How do I use Astro ASO data through MCP?

Install Tim Broddin's astro-mcp-server, a community-built bridge that reads the SQLite database the Astro Mac app stores locally (source: github.com/TimBroddin/astro-mcp-server). It exposes four read-only tools — search_rankings, get_historical_rankings, list_apps, and get_app_keywords — so you can query your tracked keyword rankings, difficulty scores, and Apple Search Ads popularity from Claude or Cursor. The server requires macOS with an active Astro subscription ($9/month) and only covers iOS. It cannot discover new keywords — it surfaces data for apps and keywords you already track in Astro. If you need Android coverage or computed difficulty scores, other servers in this comparison fill those gaps.

Is a hosted ASO MCP server better than a local one?

It depends on your use case. A hosted server — Sonar's, or the remote ones from AppTweak, Sensor Tower and AppFollow — eliminates rate-limit risk from Apple and Google, works identically across all MCP clients without local setup, and returns consistent data because it queries a maintained index rather than scraping on demand. Local servers are free and give you full control, but they depend on store availability — Apple 503 errors and Google throttling can break queries mid-session. For production automation pipelines, hosted servers are more reliable. For occasional manual research, local scrapers work fine.

Ready to add keyword difficulty, popularity, and competitor data to your AI workflow? Try Sonar's MCP server free — 50 credits on signup, no subscription required.