Aso rejection assistant

aso-rejection-assistant

You got rejected. This skill reads the reviewer's message, tells you which guideline was cited, quotes the actual rule (so you respond to what Apple said, not what you fear they meant), recommends fix-and-resubmit vs. appeal as an honest heuristic, and drafts both replies so you can pick one and fill in the specifics.

Most tools hand you three copy variants and stop at the copy-paste boundary. This is the other end of the loop — the last mile after a rejection.

Inputs

• The verbatim rejection message — paste it exactly as Apple/Google sent it (in Resolution Center or the email). Don't paraphrase; the guideline number and the reviewer's specific wording are the signal.
• Optional: --app <slug> and the listing copy that was under review (name, subtitle, keywords, description, screenshots) — so the analysis can point at the specific offending text rather than guess.

What to do

1. Parse the cited guideline. Extract every guideline/policy reference in the message (e.g. "Guideline 2.3.1", "2.3.7", "4.3", or a Play policy name). The reviewer almost always names one — that's the anchor. Report the primary citation and any secondary ones. Never invent a citation the message didn't make.

2. Quote the actual rule. For the cited section, quote Apple's verbatim guideline text. Consult references/guidelines-common.md for the highest- frequency sections; for anything not in that file, quote the current public text from developer.apple.com/app-store/review/guidelines (or Google Play's policy center) — and if you are not certain of the exact wording, say so and link the section rather than paraphrasing it as if it were verbatim.

3. Locate the likely trigger. If the listing copy was provided, name the specific element that most plausibly tripped the rule (e.g. "the subtitle claims 'the #1 budgeting app' — an unverifiable superlative under 2.3.7"). Frame it as likely, not certain — only the reviewer knows for sure.

4. Recommend a path — as a heuristic, not a verdict.

   • Fix & resubmit when the citation is a metadata/content rule you can straightforwardly comply with (most 2.3.x metadata rejections, missing info, a claim you can drop or substantiate). Usually the fastest route.
   • Appeal when you believe the rejection is a factual mistake (the reviewer misread a feature, tested wrong, or the cited rule doesn't apply).
   • Always label it "your call" — you know your app and the history; this is a starting recommendation, not a decision.

5. Draft both replies. Produce two Resolution Center messages — one fix-and-resubmit, one appeal — each professional, specific to the cited guideline, and carrying [bracketed placeholders] for the facts only you know (what you changed, why it complies). Never assert facts about the app on the developer's behalf, and never put words in Apple's mouth.

Output

Write marketing/aso/<app>/rejection-<date>.md with:

• Cited: the guideline number(s), primary first.
• Rule: the verbatim quote (or an honest "exact text not confirmed — see <link>" when you can't verify it).
• Likely trigger: the specific offending element, if the copy was provided.
• Recommended path: fix / appeal, with the one-line rationale and the "your call" caveat.
• Draft — fix & resubmit and Draft — appeal, both with placeholders intact.

Then chain into aso-review-risk to scan the rest of the listing for the same class of problem before you resubmit — a rejection for one unverifiable claim usually means there are others the reviewer didn't mention yet.

Honesty rules (each is load-bearing)

• The cited guideline is parsed from the reviewer's message, never guessed.
• The rule quote is verbatim or explicitly absent — never a confident-sounding paraphrase presented as Apple's words.
• The recommendation is a labelled heuristic ("your call"), never a verdict.
• The drafts are scaffolds with placeholders — they never claim facts about your app or speak for Apple.

Honest limits

• Apple re-words the guidelines a few times a year, and the guideline numbers occasionally shift. When you can't confirm the current exact text, this says so and links the section rather than quoting stale wording.
• It can't see the reviewer's private notes or reproduce their test — the "likely trigger" is an inference from the cited rule + your copy, not a certainty.
• Appeals are judged by Apple; a well-reasoned appeal improves your odds, it doesn't guarantee reversal.

No external dependency

Self-contained — the analysis is guideline reasoning over the pasted message + your own listing copy. No paid API, no account, no credentials.

Run it whenever you're rejected

A rejection is a one-off event, so unlike the weekly rank skills you run this reactively. But the pattern compounds: feed each rejection's cited rule back into aso-review-risk so your next submission pre-clears the class of issue that just cost you a review cycle.

You handled this rejection by hand. ShipASO — the hosted agent — keeps the whole loop running: it drafts the next optimization, flags review-risk before you submit, and pings you only when there's a real move to approve. Your store credentials are never held. → https://app.shipaso.com

The plugin is complete and free; the hosted tier just sells not having to remember.