Skip to main content

App Store Submission Checklist for Claude Code and Codex Projects

By the OnlineAppCreator editorial team4 min read

Short answer

Before submitting an app built with Claude Code or Codex, check five areas: the code (no placeholders, debug settings or development endpoints), the build (signing, version numbers, icons, release configuration), the listing (screenshots, description, URLs), privacy (manifest, labels, Data safety form, account deletion) and the review notes (demo account and explanations). Most first-time rejections of agent-built apps come from these areas, not from the core features.

On this page
  1. How to use this checklist with your agent
  2. The checklist
  3. The items agents most often get wrong
  4. Usability is part of review
  5. After you press submit

Coding agents produce working features quickly, and they also leave a characteristic set of loose ends: a test screen still reachable from the menu, a TODO string in a settings label, an API pointing at staging, a permission requested without a reason. Human reviewers at Apple and Google spot these in minutes. This checklist is the one we use before an agent-built app goes to review. It works for native, cross-platform and wrapped web apps alike, and most items can be checked by the agent itself if you ask precisely.

How to use this checklist with your agent

Give the list to Claude Code or Codex as a task: "Go through each item, check the repository and the release build, and report pass, fail or needs human check with evidence." The agent can verify code-level items reliably, such as searching for placeholder text, hard-coded URLs and missing purpose strings. Items that depend on the App Store Connect or Play Console state, legal content or judgment remain with a person. Keep the report in the repository so the next release starts from it.

The checklist

  1. No placeholder content. Search for lorem ipsum, TODO, FIXME, "test", sample names and dummy images in user-facing strings and assets.
  2. No debug features in release. Debug menus, feature flags set to development, verbose logging of user data and test payment modes are disabled or removed.
  3. Production endpoints. The release configuration points at the production backend, and that backend is reachable from outside your network, including from other countries.
  4. Every link works. Privacy policy, terms, support and any in-app links open correctly and do not lead to empty or default pages.
  5. Graceful failure. The app shows a useful message without network, after a server error and when a permission is denied. No blank screens, no infinite spinners.
  6. Account lifecycle. Sign-up, login, password reset, logout and account deletion inside the app all work. Account deletion is required by both stores if users can create accounts.
  7. Login options. If you offer third-party or social login on iOS, check Apple's current rules on offering an equivalent privacy-focused login such as Sign in with Apple.
  8. Payments. Digital goods and subscriptions use the store's in-app purchase system where required, restore purchases works, and subscription terms are shown before purchase.
  9. Identifiers and versions. Bundle ID and application ID are final, the version is set, and the build number or version code is higher than any previous upload.
  10. Signing. The release build is signed with the distribution certificate or the registered upload key, not a debug key.
  11. Icons and launch. App icons in all required sizes, no transparency in the iOS icon, a launch screen that matches the app.
  12. Platform targets. Android target API level meets Google Play's current requirement; iOS build uses an Xcode and SDK version Apple currently accepts.
  13. Permissions. Only necessary permissions are requested, each with a clear purpose string on iOS and a justification for sensitive permissions on Android.
  14. Privacy manifest and declarations. PrivacyInfo.xcprivacy is present and complete, privacy nutrition labels and the Data safety form match the code and all SDKs.
  15. Store listing. Screenshots show the real current app on the required device sizes, the description matches actual features, and no other platform or price claims appear that the store does not allow.
  16. Age rating and content. Questionnaires answered truthfully, including for user-generated content and AI-generated output, with reporting and blocking features where required.
  17. Review notes. A working demo account with sample data, an explanation of features that are not obvious, and contact details of someone who answers quickly.
  18. Native value for wrapped apps. If the app is a web app in a native shell, it offers app-specific features such as push notifications, biometric login and an offline screen, and does not show web-only elements.

The items agents most often get wrong

Three items fail more often than the rest in agent-built projects. Item 3, because agents create environment switches for convenience and do not always wire the release configuration correctly. Item 13, because agents add permissions to make a feature work and write generic purpose strings like "This app needs camera access", which Apple may reject as insufficiently specific. And item 14, because SDKs added in the middle of development change the privacy picture without anyone updating the declarations; our guide on privacy manifests and Data safety forms describes an inventory method for this.

Usability is part of review

Reviewers judge whether an app is complete and useful, and that judgment is partly about experience: confusing navigation, tiny tap targets, unreadable text or a broken layout on a small phone can push a borderline app toward rejection, and they certainly push users away. A structured outside review helps. uxreviewer.io provides expert UX and usability reviews of apps with concrete action steps that you can hand directly to your coding agent.

After you press submit

Monitor the review status and respond to messages the same day. If the app is rejected, read the cited guideline carefully, fix the cause rather than the symptom, and explain the change in your reply. Planning ranges for each step are in App Store and Google Play review times, and the most common causes are covered in top App Store rejection reasons for AI-built apps.

If you want experienced people to run this checklist and the submission with you, AppSubmitter.io provides hands-on assistance for the App Store and Google Play, including listings, privacy forms, reviewer communication and rejections, until the app is live. For teams that build and launch digital products under time pressure, that support is often the fastest way to turn a finished Claude Code or Codex project into a live app.

Frequently asked questions

Can Claude Code or Codex run this checklist automatically?

They can check most code and build items, such as placeholder text, endpoints, permissions and version numbers. Store console settings, legal texts and judgment calls need a person.

Is account deletion really required?

Yes. Both Apple and Google require apps that allow account creation to let users delete their account, and Google also requires a web-based deletion request option.

What should go into the review notes?

A working demo account with realistic data, short explanations of non-obvious features or hardware needs, and a contact who can respond quickly to reviewer questions.

Want this done right, the first time?

OnlineAppCreator takes products from idea to operation. Tell us where you stand and get a concrete plan in a free 15-minute call.