Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Expo Release Preflight

    1

    A failed production build costs 15–40 minutes of queue and build time.

    Free

    Security scanned
    Expo Release Preflight

    Expo Release Preflight

    Example session with this skill installed

    I'm cutting the production build tonight. Anything that'll bite me?

    • Read your context and instructions
    • Compiled the expo release preflight

    Release preflight: clean 1.0.0 (production)

    Verdict: Don't build tonight as-is. Three problems would ship a broken or leaky build, and one would likely fail the build itself.

    Must fix first

    The production profile builds a dev client. eas.json → build.production has "developmentClient": true. That build is a dev launcher, not your app, and the stores will reject it. Delete that line.

    1. Your OpenAI key would be readable by anyone who downloads the app.

      • src/api.ts:2 reads EXPO_PUBLIC_OPENAI_KEY. Anything prefixed EXPO_PUBLIC_ gets copied into the JS bundle in plain text, where anyone can pull it out.
      • Worse, .env holds what looks like a live sk-pro… key, and it's committed to git (commit 80498ab). Your .gitignore only ignores .env*.local.
      • There's no git remote, so it hasn't been pushed anywhere, which is good. Still:
        • Rotate the key now.
        • Untrack the file: git rm --cached .env, then add .env to .gitignore.
        • Move the OpenAI call behind your own backend. A key inside the app can't be kept secret.

    The app calls http://localhost:3000. src/api.ts:1 hardcodes it, so fetchNotes() fails on every real device. iOS also blocks plain http by default. Read the URL from something like EXPO_PUBLIC_API_URL, and set it in the EAS production environment: bunx/npx eas-cli env:create… or the current eas env:set --name EXPO_PUBLIC_API_URL --value https://… --environment production.

    1. Google Sign-In will probably break the build or crash at login.
      • app.json lists "@react-native-google-signin/google-signin" as a plugin with no options. Without Firebase it needs an iosUrlScheme (your reversed iOS client ID), and iOS prebuild errors without it.
      • src/auth.ts calls GoogleSignin.configure() without a webClientId or iosClientId, so sign-in will fail at runtime even if the build succeeds.

    Check before building

    • [ ]

    Camera permission text is empty. app.json sets ios.infoPlist.NSCameraUsageDescription: "". Apple rejects an empty string. Remove the line and add ["expo-camera", { "cameraPermission": "Allow $(PRODUCT_NAME) to scan your notes." }] to plugins.

    • [ ]

    Version number. A v1.0.0 tag already exists. If 1.0.0 has already gone to the stores, bump version in app.json. With appVersionSource: "remote", EAS only bumps build numbers, never this.

    • [ ]

    Set the environment explicitly. Add "environment": "production" to build.production. Then confirm the variables with eas env:list --environment production, because your local .env doesn't reach EAS.

    • Run npx expo-doctor. I skipped it. Check that the expo-camera and google-signin versions match SDK 58.
    • Run lint and typecheck: npx expo lint, npx tsc --noEmit.

    Will bite later

    Updates: the production profile has no channel, so this build can never receive an EAS Update hotfix. Add "channel": "production" now if you want that option.

    • Crash reporting: there isn't any (no Sentry, Crashlytics or similar), so you won't see crashes from real users.

    App.tsx is still the template screen. ScanNote, signIn and the notes API aren't rendered anywhere. Make sure the build actually contains the features you mean to ship.

    submit.production is empty. --auto-submit will prompt for, or fail without, App Store Connect and Play credentials.

    Smoke test, on a release build on a real device

    • Cold start, Google login, fetching notes from the real API, an AI summary going through your backend, camera scan and its permission prompt
    • Offline start
    • If 1.0.0 is already in the stores, install this version over it

    Next step

    Fix items 1–4 and the permission text, then do an internal release build (eas build --profile preview) on a device before running eas build -p all --profile production. I can make the config fixes and untrack .env if you'd like. Rotating the key and moving the OpenAI call to a server are up to you.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A failed production build costs 15–40 minutes of queue and build time. A successful build with the wrong API URL, a missing env var or stale native config costs a whole store review cycle. This skill catches both before eas build runs.

    What it does

    • Static release check (scripts/release-check.mjs, zero-dependency Node 18+): reads app.json / app.config, eas.json, package.json, git state and your source.
      • Production profiles that build a dev client, or don't set an EAS environment
      • EXPO_PUBLIC_ variables your code reads that the EAS environment doesn't define, and ones that look like secrets
      • localhost / LAN URLs and leaked keys in app code
      • gitignored assets that EAS won't upload, and committed ios/ / android/ folders that silently ignore app.json changes
      • Versions already tagged, runtimeVersion and channel setup
      • Optional: --run-expo-checks runs expo-doctor and expo install --check
    • Release checklist and smoke-test plan for what a script can't see: migrations, upgrade installs over the previous store version, offline start, deep links.
    • Hand-offs: store-review risks go to app-store-review-guard, build failures to eas-build-submit, JS-only fixes to expo-ota-updates.

    Output

    A release preflight with a verdict, what must be fixed before building, what to check, what will bite later, and a smoke-test list. The script prints a human report or --json and exits non-zero on blockers, so it also works in CI.

    Open source (MIT): https://github.com/disconnectuss/expo-ship-kit.skills. Part of expo-ship-kit.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Mobile Development.

    Static checks only: it can't run your app, test your backend or see EAS-side settings beyond what eas.json and eas env:list show. expo-doctor runs only with --run-expo-checks and needs installed node_modules. The script needs Node.js 18+ and git, and runs from the Expo project root.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 day ago

    • Free with an Agensi account

    Software
    Node.jsexpoGitexpo-doctor
    Optional
    expo-doctor via npx (only with --run-expo-checks)

    Permissions

    Terminal / Shell
    Read Files

    Terminal: runs `node scripts/release-check.mjs`, which reads the local Expo project and calls git and `npx --no-install expo config --json`. Only with the opt-in `--run-expo-checks` flag does it run `npx expo-doctor@latest` and `npx expo install --check`, which download from npm. Read Files: app config, eas.json, package.json, .env files and source, to find leaked keys and local URLs. Nothing is sent anywhere. The key-like regexes in the script are leak detectors, not secrets.

    Listed1 day ago

    What's inside

    Frequently Asked Questions

    expo-doctor checks dependency versions and known config problems. This skill checks whether the release build will contain what you think: the EAS profile and environment, EXPO_PUBLIC_ variables, local URLs, leaked keys, gitignored assets, committed native folders, versions and channels. It can also run expo-doctor for you with --run-expo-checks.

    Any agent that reads SKILL.md: Claude Code, Codex CLI, Cursor, Gemini CLI and others. To run the script the agent needs shell access to your project; without it, the skill still walks the checklist by reading your files.

    SKILL.md (the release workflow and report format), scripts/release-check.mjs (zero-dependency Node 18+ check, human or --json output) and references/release-checklist.md. Rules were checked against Expo SDK 58 and the EAS docs on 29 Oct 2026. It's free and MIT-licensed; the same files are on GitHub.

    Yes, by detecting committed 'ios' or 'android' folders, the skill identifies if your project is ignoring app.json changes, which is the most common cause of configuration drift in managed Expo projects.

    Node.js 18+ and git, run from the Expo project root. The EAS CLI isn't needed for the check itself; the report suggests `eas` commands (like `eas env:list`) for you to run.

    No, the skill performs static analysis of your configuration and environment variables; it does not require an active internet connection to EAS or the App Store to run its core audits.

    New in Mobile Development

    Free