tools / reviews

reviews

read-only

Recent customer reviews

What are users saying — in the words they'd also search with?

reviews returns an app's most recent customer reviews per storefront from Apple's public RSS feed — pages 1–10, roughly 50 reviews per page — with the rating distribution and average of the fetched page alongside the texts.

For ASO, reviews are a vocabulary mine: users describe the job in their own words (“streak”, “accountability”, “lifetime subscription”), and those words are keyword candidates your metadata may not serve yet. Run them on your competitors and you get their users' complaints as your positioning input.

The distribution describes the fetched page, not the app's lifetime rating — recent sentiment, which is exactly what a release just changed. Compare against the aggregate rating from app.

worked example

The 3 most recent US reviews for HabitKit, with the page's rating distribution.

tools/call → reviews

{
  "name": "reviews",
  "arguments": {
    "appId": 6443918070,
    "countries": [
      "US"
    ],
    "limit": 3
  }
}

reviews → structuredContent (excerpt)

{
  "appId": 6443918070,
  "page": 1,
  "storefronts": [
    {
      "country": "US",
      "count": 50,
      "averageRating": 4.52,
      "distribution": { "1": 2, "2": 2, "3": 1, "4": 8, "5": 37 },
      "reviews": [
        { "rating": 5, "title": "This is my favourite habit tracking app",
          "text": "Been using this for over 6 years now. … Simple UI. Elegant and efficient.",
          "version": "1.15.2", "date": "2026-07-27T09:31:04-07:00" },
        { "rating": 5, "title": "It's really Good", … },
        { "rating": 5, "title": "Can't find the lifetime version",
          "text": "… I wanna buy the lifetime subscription but I can't find it anywhere …",
          "version": "1.15.2", "date": "2026-07-10T17:49:21-07:00" }
      ]
    }
  ]
}

real response, captured 2026-07-29 — trimmed only where marked with …

The third review is a five-star complaint — a purchase blocker (“can't find the lifetime version”) that never shows up in the rating alone.
distribution covers the ~50 fetched reviews: recent sentiment, so a bad release shows here days before it dents the lifetime average.
Review language is keyword research: “habit tracking app”, “lifetime subscription” — phrases to score with volume.

when to use it

Post-release monitoring: the fetched-page average reacts to a bad build in days, the all-time rating takes months.
Vocabulary mining for metadata — yours and your competitors' — before an audit-driven rewrite.
Per-market sentiment: the same version can hold 4.8 in the US and 3.9 in Germany; reviews are per storefront like everything else here.
Support radar for agents: recurring phrases in low-star reviews are a triage queue.

description — as served to agents

Recent customer reviews for an app from Apple's public RSS feed, per storefront: rating distribution, average of the fetched reviews, and review texts (for mining user phrasing as keyword candidates). Pages 1-10, ~50 most-recent reviews per page.

input

appId *

integer

App Store trackId of the app

countries

string[]

2-letter storefront codes, e.g. ["US", "GB", "JP"]. Each storefront is queried separately; at most 25 per call.(default ["US"])

page

integer

RSS feed page(default 1)

limit

integer

Max review texts included per storefront(default 10)

output

appId *

integer

page *

integer

storefronts *

object[]

country *

string

count *

integer

reviews fetched from this page

averageRating *

number | null

distribution *

object

1 *

integer

2 *

integer

3 *

integer

4 *

integer

5 *

integer

reviews *

object[]

id *

string

rating *

integer

title *

string

text *

string

author *

string

version *

string | null

date *

string | null

note *

string

pairs with

appcompare recent sentiment against the lifetime rating›competitorswhose reviews to mine next›volumescore the phrases users actually wrote›

endpoint https://mcp.openaso.ai/mcp · auth Authorization: Bearer <key> · get a key