tools / coverage

coverage

read-only

Metadata keyword coverage

Which searches can this metadata draft actually rank for?

Apple indexes a localization's title (30 chars), subtitle (30) and hidden keyword field (100) as one combined word pool: an app can rank for any phrase assembled from words across those fields. coverage operationalizes that model, locally and instantly — no network, no quota.

Give it a draft and target phrases, and it returns the pool with strongest-field attribution, cross-field duplicates (pure reclaimable waste), per-field character budgets, keyword-field waste accounting (commas count; spaces around commas don't have to), and a verdict per phrase: covered verbatim, covered cross-field, or missing specific words.

Weight phrases with {phrase, value} — typically popularity from volume — and wordValues aggregates each word's total search value across every phrase it serves, with unlockValue (what adding an absent word would newly cover) and atRiskValue (what dropping a word would break) as the marginal numbers character-budget decisions should be made on.

worked example

Evaluate HabitKit's live title and subtitle plus a drafted keyword field against weighted target phrases.

tools/call → coverage

{
  "name": "coverage",
  "arguments": {
    "title": "Habit Tracker - HabitKit",
    "subtitle": "Streaks & Accountability",
    "keywords": "daily, goals,routine,streak,habits,widget,grid,minimal",
    "phrases": [
      {
        "phrase": "habit tracker",
        "value": 82
      },
      {
        "phrase": "streak tracker",
        "value": 76
      },
      "routine planner",
      "habit grid"
    ]
  }
}

coverage → structuredContent (excerpt)

{
  "fields": [
    { "field": "title",    "chars": 24, "limit": 30,  "remaining": 6 },
    { "field": "subtitle", "chars": 24, "limit": 30,  "remaining": 6 },
    { "field": "keywords", "chars": 54, "limit": 100, "remaining": 46 }
  ],
  "phrases": [
    { "phrase": "habit tracker", "covered": true, "exactIn": "title", "value": 82 },
    { "phrase": "streak tracker", "covered": true, "exactIn": null, "crossField": true,
      "words": [ { "word": "streak", "match": "plural", "field": "subtitle" },
                 { "word": "tracker", "match": "exact", "field": "title" } ], "value": 76 },
    { "phrase": "routine planner", "covered": false, "missing": ["planner"] },
    { "phrase": "habit grid", "covered": true, "crossField": true }
  ],
  "keywordFieldWaste": { "spaces": 1, "duplicates": 0, "stopwords": 0, "total": 1 },
  "wordValues": [
    { "word": "tracker", "inPool": "title", "phrases": ["habit tracker", "streak tracker"],
      "totalValue": 158, "unlockValue": 0, "atRiskValue": 158 },
    { "word": "planner", "inPool": null, "chars": 7, "phrases": ["routine planner"],
      "totalValue": 1, "unlockValue": 1, "atRiskValue": 0 },
    …
  ]
}

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

streak tracker (demand 76) is covered without appearing anywhere verbatim: plural Streaks in the subtitle assembles with tracker from the title. That is the cross-field model most ASO tooling misses.
keywordFieldWaste.spaces: 1 caught the space in daily, goals — one character of the 100 doing nothing.
tracker carries atRiskValue: 158 — drop it and both weighted phrases break. planner costs 7 chars to unlock routine planner; with 46 keyword-field characters remaining, that is an easy add.

when to use it

Drafting metadata: iterate title/subtitle/keyword-field candidates locally until target coverage is total and waste is zero — then verify with live rank calls.
Regression-checking a rewrite: pass every phrase you currently rank for; any that comes back covered: false is a ranking you are about to silently give up.
Budgeting the 100-char keyword field: waste accounting plus unlockValue tells you exactly which 7 characters earn their place.
Understanding a competitor's reach: their title + subtitle (from app) is two-thirds of their pool — see which phrases it already assembles.

description — as served to agents

Which search phrases a metadata draft can rank for, computed locally (no network). Apple indexes the title (30 chars), subtitle (30) and hidden keyword field (100) of a localization as ONE combined word pool: an app can rank for any query assembled from words across those fields — "photo" in the title plus "resize" in the keyword field can rank for "resize photo". Title words weigh most, then subtitle, then keyword field; word order and case do not gate what can rank (a phrase kept intact in the title/subtitle is a mild extra signal); repeating a word in a second field adds nothing. Reports per-field character budgets, the combined pool, cross-field duplicates (reclaimable characters), and for each target phrase whether it is covered — verbatim in the title/subtitle, assembled cross-field, or missing words. In the keyword field, commas count toward the 100 characters; redundant characters (spaces around commas, duplicated words, stopwords) are reported in keywordFieldWaste. Weight phrases with {phrase, value} (e.g. value = popularity from the volume tool) and wordValues aggregates each word's total search value across every phrase it combines into — "photo" serves "photo editor", "photo maker" and "take photo" at once. totalValue credits a phrase to all of its words (not additive across words); use unlockValue (absent words whose addition alone covers new phrases) and atRiskValue (pool words whose removal breaks covered phrases) for marginal decisions. Evaluates one localization; storefronts may index extra localizations (e.g. the US storefront also indexes Spanish (Mexico)). Cross-field combinations tend to rank weaker than same-field phrases, so confirm real positions with rank.

input

title *

string

App title draft (30-char limit)

subtitle

string

Subtitle draft (30-char limit)

keywords

string

Hidden keyword field draft, comma-separated (100-char limit; commas count, spaces are waste)

phrases

object[]

Target search phrases to check coverage for; weight with {phrase, value} to aggregate per-word search value in wordValues

output

fields *

object[]

field *

"title" | "subtitle" | "keywords"

text *

string

chars *

integer

limit *

integer

Apple's limit: title 30, subtitle 30, keywords 100

remaining *

integer

characters left; negative = over the limit

overLimit *

boolean

tokens *

string[]

pool *

object[]

combined indexable word pool, attributed to the strongest field containing each word

word *

string

field *

"title" | "subtitle" | "keywords"

strongest field containing the word

stopword *

boolean

filler word Apple's search largely ignores — reclaimable

alsoIn *

("title" | "subtitle" | "keywords")[]

weaker fields repeating the word (reclaimable characters)

duplicates *

object[]

words spent in more than one field — repeating adds no ranking benefit

word *

string

fields *

("title" | "subtitle" | "keywords")[]

phrases *

object[]

phrase *

string

covered *

boolean

every substantive word is available in the combined pool (stopwords assumed ignored; false when the phrase has no substantive match at all)

exactIn *

"title" | "subtitle" | null

whole phrase appears verbatim in this field (never keywords — that field is unordered); a kept-intact phrase is a mild extra relevance signal

crossField *

boolean

covered only by combining substantive words from different fields (ranks, but tends to rank weaker)

words *

object[]

word *

string

match *

"exact" | "plural" | "stopword" | null

"plural" is a suffix heuristic; null when missing

field *

"title" | "subtitle" | "keywords" | null

strongest field supplying the match; null for stopword/missing

missing *

string[]

words in no field — the pool alone cannot produce the phrase (Apple can still rank it via developer name, in-app purchase names or query expansion)

value *

number

the phrase's relative search value (input weight; 1 when unweighted)

keywordFieldWaste *

object | null

keyword-field characters that deleting the waste would actually free (null when no keywords draft was given)

spaces *

integer

whitespace around commas or at the ends — a space that is a word's only separator is char-neutral and not counted

duplicates *

integer

chars of words already in title/subtitle or repeated in the field

stopwords *

integer

chars of filler words

separators *

integer

commas/spaces freed by removing the wasted words

total *

integer

wordValues *

object[]

per-word aggregate search value across all target phrases, sorted by totalValue. A word serves every phrase it can combine into, but a phrase's value is credited in full to EACH of its words — totals are not additive across a word set; budget with unlockValue/atRiskValue

word *

string

inPool *

"title" | "subtitle" | "keywords" | null

strongest field containing — or plural-matching — the word; null if absent

chars *

integer

characters of the word itself; a keyword-field entry additionally costs one separating comma

phrases *

string[]

totalValue *

number

summed value of every target phrase containing the word

unlockValue *

number

absent words: summed value of phrases covered by adding just this word

atRiskValue *

number

in-pool words: summed value of covered phrases that would actually lose coverage if the word were dropped from every field (a surviving plural variant keeps a phrase safe)

note *

string

pairs with

volumephrase weights should be measured demand, not guesses›rankcoverage says possible; rank says actual›appfetch the live title + subtitle you are evaluating›

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