◉RORK LABJP
●MCP — Rork now lets you add and manage your own MCP servers inside Plugins (11/21)●2.5 ROUTE — From 12/1, apps calling gemini-2.5-flash are routed to 3.5 Flash-Lite and billed at its rates (9 days left)●CHANNEL — Users report Android notification channel reads and writes failing in Expo Go●NEW — New: three clues to check before replying to a stuck-on-splash review●EXPORT — Export your projects before you downgrade or cancel a plan●SONNET 4.5 — Apps that call Sonnet 4.5 through the API should switch before it retires on 11/30●MCP — Rork now lets you add and manage your own MCP servers inside Plugins (11/21)●2.5 ROUTE — From 12/1, apps calling gemini-2.5-flash are routed to 3.5 Flash-Lite and billed at its rates (9 days left)●CHANNEL — Users report Android notification channel reads and writes failing in Expo Go●NEW — New: three clues to check before replying to a stuck-on-splash review●EXPORT — Export your projects before you downgrade or cancel a plan●SONNET 4.5 — Apps that call Sonnet 4.5 through the API should switch before it retires on 11/30
Articles/Dev Tools
⬡ Dev Tools/2026-08-19Intermediate

Three conditions that make Play Policy Insights report nothing on an Expo project

Google Play now ships an open-source policy auditing skill. Running it against an Expo-shaped project, one directory argument moved the result from five detected data categories to zero. Here is when the scan actually reaches your code, and where its output should not be trusted.

Google Play44Expo237Rork620Data safetyApp review

✦ Premium Article

A clean report is the result I trust least.

In its July 15, 2026 policy announcement, Google Play pointed developers at an open-source auditing tool: something that grounds an LLM in Play's policy documents so your assistant can evaluate your code from inside your IDE or CLI. It is called Play Policy Insights. Anything that reduces the number of rejection round-trips is worth an afternoon.

So I built a minimal Expo-shaped project and ran it. First pass, pointed at android/app: zero findings. Second pass, pointed at the repository root: five data collection categories.

Same code. Same permissions. Same dependencies. The only difference was the path I typed.

Update (October 28, 2026): The target API level 36 requirement discussed below took effect on August 31, 2026, so it is no longer a deadline to beat. The wording now reflects that. The findings about scan entry points are unchanged from publication.

What the skill actually inspects

The skill lives in the android/skills repository. It audits three policy domains: Permissions and APIs Hygiene, User Account and Identity, and Data Safety and Privacy.

It runs in two phases. Phase 1 executes orchestrator.py init <app_dir>, which performs static analysis, maps the codebase, and decides which audit goals to activate. Phase 2 hands each activated goal to an agent as a generated prompt, collects the JSON results, and generate_report.py turns them into a Markdown compliance report.

PhaseCommandOutput
Phase 1orchestrator.py init <app_dir>activated_goals, static analysis JSON, per-goal prompts
Phase 2Run per-goal prompts, then orchestrator.py aggregateaggregated_findings.json
Finalizegenerate_report.py <temp_dir>compliance_report.md

Everything rests on Phase 1. Whatever the static scan misses is gone — no downstream model recovers it. The skill says as much: the Phase 1 audit is the source of truth, and if orchestrator.py fails you must stop rather than fall back to manual review. Which means the dangerous case is not a loud failure. It is Phase 1 quietly scanning nothing.

Condition 1: the entry point has to be the repository root

scanner.py checks the directory you gave it and its parent for a package.json or pubspec.yaml. If it finds one, it flags the project as hybrid and widens the scan root to that directory.

In an Expo project, package.json sits at the repository root. Point the tool at android/app and the parent is android, which has no package.json. Hybrid detection never fires, and not a single line of your JavaScript or TypeScript is read.

Directory passed to initactivated_goalsData categories detected
Repository rootdata_safety_part_1, permissions_and_apis5
android/app(empty)0

When a tool says "Android app audit," pointing at the android directory feels correct. That is what I did first. But in Expo and React Native the code requesting permissions lives on the JavaScript side. Scan only the native tree and the audit stops being an audit.

The tell is one line on stderr: Hybrid app detected. Expanding scan root to: .... If that line is absent, the report is not worth reading.

Scanning stderr by eye is how I miss it, so I send it to its own file and let a script do the looking.

# Entry check: stop immediately if the hybrid-app line never appeared
python3 orchestrator.py init . 2> init.err > init.json
grep -q 'Hybrid app detected' init.err \
  || { echo "NG: no hybrid detection. Check your scan root." >&2; exit 1; }

Splitting stdout from stderr is only two lines, but it takes the job of noticing an absence away from tired eyes.

✦

Thank you for reading this far.

Continue Reading

What follows includes implementation code, benchmarks, and practical content we hope you'll find useful. This site runs without ads — server and development costs are supported entirely by members like you. If it's been helpful, we'd be truly grateful for your support.

WHAT YOU'LL LEARN
✦You will be able to tell the difference between a pre-submission check that found nothing and one that never reached your code in the first place
✦You will be able to point the official skill at the right entry point in an Expo or React Native project and close Data safety gaps before you submit
✦You will be able to see where static scanning over-reports and where it is structurally blind, so you never copy a machine-generated finding straight into your Play Console declaration
Secure payment via Stripe · Cancel anytime
✦

Unlock This Article

Get full access to the rest of this article. Buy once, read anytime. This site is ad-free — your support goes directly toward keeping it running.

or
Unlock all articles with Membership →
Share

Thank You for Reading

Rork Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.

  • ✦Copy-paste ready implementation code
  • ✦New advanced guides published daily
  • ✦$5/mo or $10 for lifetime access
View Membership →

Related Articles

⬡ Dev Tools2026-11-19
New Wallpapers Showed Up Everywhere Except the Filters: Ship the Count and the Category Lists Together
I added 67 wallpapers in one batch. They appeared in the full list and in New, but not under a single category filter, because I had raised the ID limit and forgotten to ship the category lists. Here is a pre-bump check, a client-side way to spot silent fallbacks in an Expo app, and a post-release count against the origin.
⬡ Dev Tools2026-10-30
It runs in development but the iOS Release build dies on launch — checking whether a missing scheme is the cause
In an expo-router app, an app.json with no scheme can crash only the iOS Release build at launch. Why Debug and Android survive, the two ways to fix it, how to think about leaving the scheme out on purpose, and a pre-submission check.
⬡ Dev Tools2026-10-15
When EXPO_PUBLIC_ Values Read Fine in Dev but Come Back undefined in Production: Telling the Spellings Apart
The API URL reads fine while developing and turns undefined once the app is built. The cause was how process.env was written. I lined up babel-preset-expo's output for dev and production and tabulated which spellings get replaced and which are left behind.