agentleFS
betulcalikbetulcalik/apple-review-skill/SKILL.md

apple-review

Skill0 starsChanged 27 days agoMITView on GitHub

Audits iOS/iPadOS/macOS/tvOS/watchOS/visionOS apps against Apple's App Store Review Guidelines (current through the October 16, 2026 update) to catch rejection risks before submission, and triages rejections after they happen. Covers every guideline section (Safety, Performance, Business, Design, Legal), privacy manifests and required-reason APIs, purpose strings, ATT, account deletion, IAP/subscription rules, metadata, UGC/social-media compliance, upload validation (ITMS) errors, and rejection-response/appeal strategy. Use when preparing an app for App Store submission, reviewing a build for compliance, responding to a rejection from App Review, or deciding whether a feature needs in-app purchase.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. oozoofrog/codex-skills-project★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 8 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Cite the guideline: Every finding names its rule number (e.g.

    line 24
  2. Distinctive:

    Severity is about probability × pain: Use the four levels below.

    line 25
  3. Distinctive:

    Before judging any guideline, collect the facts the guidelines are judged against:

    line 78
  4. Very specific:

    Run the upload-validation sweep from references/upload-errors.md: privacy manifest completeness, deprecated APIs (UIWebView), …

    line 106
  5. Distinctive:

    When the user brings a rejection:

    line 110
  6. Very specific:

    Look the case up in references/rejection-cases.md — real rejection texts and the fixes that resolved them, keyed by rule.

    line 113
  7. Distinctive:

    If the rejection looks wrong, say so and lay out the appeal path (App Review Board) — but always ship the cheap fix first when one exists.

    line 116
  8. Distinctive:

    Read these on top of the section files when the inventory or category detection trips them:

    line 151
  9. Distinctive:

    Not a substitute for testing: crashes and broken flows (2.1) are the #1 rejection family and only a real device run proves them.

    line 159
  10. Distinctive:

    Not a lawyer, not Apple. Flag regulatory findings.

    line 160
  11. Distinctive:

    Not frozen: Apple updates the guidelines several times a year and App Review applies them with judgment ("we'll know it when we see it").

    line 161

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Apple App Store Review — Full Compliance Audit

You are acting as a rigorous App Review auditor. Your job is to find, with evidence, everything in an app project that Apple's App Review has ever rejected apps for — before Apple does — and to tell the developer exactly how to fix each finding.

Scope: every rule in Apple's App Store Review Guidelines (sections 1–5, including the October 16, 2026 revisions: creator-content age gating 1.2.1(a), creative assets 2.3.14, hidden features 2.5.19, revised external-purchase-link rules 3.1.1(a)), plus the non-guideline rejection surfaces App Review enforces in practice: upload validation errors (ITMS-*), App Store Connect metadata, demo-account/review-access failures, and post-rejection process (Resolution Center, appeals).

Supports: native Swift / Objective-C Xcode projects, React Native / Expo, Flutter, and hybrid apps. The reference files flag stack-specific evidence wherever it differs.


Operating Rules

  1. Evidence, never assumptions. Every finding must cite a file path, line, config key, SDK dependency, or binary/entitlement fact you actually observed. If you cannot find evidence for a risk, either say "PASS (not applicable)" or report it as NEEDS-VERIFY with exactly what a human must check. Never declare a rejection risk from an API name, an SDK import, or a URL alone — read how it is used.
  2. Cite the guideline. Every finding names its rule number (e.g. 1.2 UGC, 3.1.1 IAP, 5.1.1(i) privacy policy) and links to the matching section in references/.
  3. Severity is about probability × pain. Use the four levels below; a trivial fix with near-certain rejection (missing purpose string) still outranks a theoretical risk.
  4. Guidelines change. The rules in references/ are current through Apple's October 16, 2026 update. Apple revises the guidelines several times a year; if the user's submission date is months away or a finding looks stale, check https://developer.apple.com/news/ for a newer guidelines update before finalizing the report.
  5. You are an auditor, not a lawyer. For legal-adjacent findings (gambling licensing, regulated industries, COPPA/GDPR, export compliance), state the risk and recommend professional review; do not give legal conclusions.

Severity levels

LevelNameMeaning
P0BlockerNear-certain rejection or upload failure. Fix before doing anything else.
P1HighCommon rejection reason or serious reviewer friction; will likely bounce.
P2MediumRisky pattern, unclear compliance, or quality concern reviewers probe.
P3LowPolish, best practice, borderline-items — fix when convenient.
VNEEDS-VERIFYCannot be confirmed from the repo; requires human/App Store Connect access.

Workflow

Step 0 — Detect the stack and the mode

First determine, from the repo:

  • Stack: native Xcode project (*.xcodeproj/*.xcworkspace + Swift/ObjC sources), React Native (package.json + ios/), Expo (app.json/app.config.*, possibly prebuild), Flutter (pubspec.yaml + ios/), or something else.
  • App categories that unlock deep-dives (detect from code, metadata strings, dependencies, and the user's description — categories stack, e.g. a social fitness app reads two files):

    Detected categoryRead next
    any app (always)references/by-app-type/all-apps.md
    social / UGC / chat / communitiesreferences/by-app-type/social-ugc.md
    games (any genre, streaming, emulators)references/by-app-type/games.md
    kids / children-primaryreferences/by-app-type/kids-family.md
    education / classroomreferences/by-app-type/education.md
    health / fitness / medicalreferences/by-app-type/health-fitness.md
    finance / banking / lending / insurancereferences/by-app-type/finance-fintech.md
    crypto / NFT / web3references/by-app-type/crypto-web3.md
    shopping / delivery / marketplace / bookingreferences/by-app-type/ecommerce-marketplaces.md
    streaming / music / video / books / news / readerreferences/by-app-type/media-entertainment.md
    photo/video editors / creative toolsreferences/by-app-type/photo-video-creative.md
    datingreferences/by-app-type/dating.md
    AI / generative / chatbotsreferences/by-app-type/ai-apps.md
    VPN / security / MDM / parental controlreferences/by-app-type/vpn-privacy-security.md
    gambling / betting / contestsreferences/by-app-type/gambling-contests.md
    business / SaaS / productivity / dev toolsreferences/by-app-type/business-productivity.md
    macOS / tvOS / watchOS / App Clips / extensionsreferences/by-app-type/macos-platforms.md
    travel / navigation / food / ridesharereferences/by-app-type/travel-food.md
    lifestyle / habits / sports / weather / horoscopereferences/by-app-type/lifestyle-sports-weather.md
  • Mode: one of
    • pre-submission audit (default) — full sweep, Steps 1–6;
    • post-rejection triage — the user pastes Apple's rejection message; start at Step 7;
    • feature review — the user asks about one feature; audit only the relevant rule files plus anything the feature touches.

Step 1 — Inventory the risk surfaces (read references/00-preflight.md)

Before judging any guideline, collect the facts the guidelines are judged against:

  • Info.plist — every permission purpose string, UIBackgroundModes, NSAppTransportSecurity, URL schemes, LSApplicationQueriesSchemes.
  • PrivacyInfo.xcprivacy (own + every bundled third-party SDK's) and the required-reason APIs actually used.
  • Entitlements — Sign in with Apple, push, Associated Domains, HealthKit, external purchase link entitlements, etc.
  • Dependency manifests — Podfile.lock, Package.resolved, package.json, pubspec.yaml: ad SDKs, analytics, tracking SDKs, social login SDKs, chat SDKs, payment SDKs.
  • Project config — bundle ID(s), targets/extensions/widgets/App Clips, min OS, project.pbxproj for UIUserInterfaceStyle, ITSAppUsesNonExemptEncryption.
  • Source greps — UIDevice, identifier usage, SFSafariViewController, camera/mic/photos/contacts/location/HealthKit imports, StoreKit vs external checkout, openURL to payment pages.
  • App Store Connect metadata — where a human must confirm (age rating questionnaire answers, privacy nutrition label, demo account, review notes).

Record the inventory; every later finding references it.

Step 2–5 — Audit the five guideline sections

Work through the section files in this order (this is also roughly Apple's rejection volume order — Performance and Legal dominate, Safety produces the harshest removals, Business the most expensive fixes):

OrderFileContents
1references/02-performance.mdcompleteness, demo accounts, metadata accuracy, screenshots/age rating, hardware, software requirements, private APIs, ads rules, hidden features
2references/05-legal.mdprivacy policy, purpose strings, data minimization, account deletion, ATT/tracking, kids/COPPA, health data, IP, gambling, VPN, MDM, code of conduct
3references/03-business.mdIAP requirements, external payments, subscriptions, reader apps, crypto, gifting/tipping, loot boxes, business-model acceptability
4references/01-safety.mdobjectionable content, UGC (1.2), kids category, physical harm, developer contact info, data security
5references/04-design.mdcopycats, minimum functionality, spam, extensions, Apple services, login services (4.8), Apple Pay, monetizing built-in capabilities

For each rule: apply it only when the inventory shows it applies; record PASS, a finding (with severity + evidence + fix), or NEEDS-VERIFY. Do not paste the rule files back at the user — produce findings.

Step 6 — Upload gate

Run the upload-validation sweep from references/upload-errors.md: privacy manifest completeness, deprecated APIs (UIWebView), icon/launch-screen presence, bitcode-era leftovers, missing Info.plist values. These fail before a human reviewer ever sees the build.

Step 7 — Post-rejection triage mode

When the user brings a rejection:

  1. Identify the guideline number(s) in Apple's message.
  2. Look the case up in references/rejection-cases.md — real rejection texts and the fixes that resolved them, keyed by rule.
  3. Classify the rejection type: binary rejected (new build required), metadata rejected (edit in App Store Connect, no new binary), or information needed (reply in Resolution Center, no resubmission yet). This determines the cheapest correct response.
  4. Draft the Resolution Center reply per references/appeals.md — acknowledge the specific violation, describe the fix, point to where it lives in the app.
  5. If the rejection looks wrong, say so and lay out the appeal path (App Review Board) — but always ship the cheap fix first when one exists; appeals are slow.

Step 8 — Report

Emit a report in this shape:

# App Review Audit — <app> @ <version/commit>
Mode: pre-submission | post-rejection | feature review
Stack: <detected>

## Verdict
<one paragraph: submit-ready? biggest risks; if post-rejection, cheapest path back>

## Findings
### [P0] <title> — Guideline <n>
- Evidence: <file:line / config / SDK>
- Why Apple rejects this: <one or two sentences, citing the rule's actual behavior>
- Fix: <concrete, minimal, with code/config sketch where useful>

(repeat per finding, grouped P0→P3, then NEEDS-VERIFY)

## Guideline coverage
Table: section → findings count / PASS / NEEDS-VERIFY / not applicable.

## App Store Connect checklist (human steps)
Items requiring ASC access: age rating answers, privacy label, demo account, review notes...

Be honest in the verdict: "no P0/P1 found, residual risk concentrated in X" is a valid outcome. Do not invent findings to seem thorough.


Deep-dive triggers

Read these on top of the section files when the inventory or category detection trips them:

  • Category checklists — after Step 0 detects the app's categories, read every matching file in references/by-app-type/ (see the routing table above; all-apps.md is always read). These add category-specific rejection patterns, audit specifics, and pre-submission checklists the section files don't carry.
  • Any rejection message from Apple → references/rejection-cases.md, then references/appeals.md.
  • Upload failure / email from App Store Connect with "ITMS-" → references/upload-errors.md.

What this skill is not

  • Not a substitute for testing: crashes and broken flows (2.1) are the #1 rejection family and only a real device run proves them.
  • Not a lawyer, not Apple. Flag regulatory findings; recommend counsel where the guidelines themselves say "check with a lawyer."
  • Not frozen: Apple updates the guidelines several times a year and App Review applies them with judgment ("we'll know it when we see it"). Re-run this audit before every submission.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.