recipes / draft-app-store-metadata

Draft App Store metadata with AI that actually ranks

AI-written metadata usually fails the same two ways: it optimizes pretty phrases when Apple matches a combined word pool, and it wastes budget on duplicates and formatting. openaso teaches your agent the real model and hands it a local verifier — so the draft ships as checked math, not vibes.

the prompt

Draft App Store metadata for <your app>, targeting these phrases:
<phrase 1>, <phrase 2>, <phrase 3>, …

Rules: title ≤ 30 chars, subtitle ≤ 30, keyword field ≤ 100. Optimize
the combined word pool, not literal phrases — no word repeated across
fields, no spaces after commas in the keyword field. Verify coverage
of every phrase and show what each word unlocks before calling it done.

what your agent does

1. 01

   app

   Pulls the current listing so the draft keeps what already works — your brand name, the words already earning ranks.

2. 02

   Drafts against the pool model: title words weigh most, then subtitle, then keyword field. A phrase can be assembled across fields, so each word is placed once, where it earns most.

3. 03

   coverage

   Checks the draft locally: budgets per field, duplicate words, keyword-field waste — commas count toward the 100, and "a, b" wastes what "a,b" doesn't — plus per-phrase coverage and per-word value.

4. 04

   rank

   After you ship, verifies assembled phrases with live ranks — cross-field combinations tend to rank weaker than same-field ones, and only real positions settle it.

what you get back

Three fields that fit their budgets, a per-phrase coverage table showing how every target assembles from the pool, and a waste report proving no character idles. When a phrase can't be covered inside the budgets, the agent says so and shows the trade — instead of quietly dropping it.

{
  "field": "keywords",
  "text": "streak,habits,routine,daily,planner,reminder",
  "chars": 45,
  "limit": 100,
  "remaining": 55,
  "overLimit": false
}

make it yours

keep going