No code execution
We do not install packages, build the app, or run scripts from your project files.
How LaunchLint reviews apps
LaunchLint connects deterministic signals in project files to reviewed store rules. A signal never becomes a guarantee of approval.
Last reviewed: July 19, 2026
We do not install packages, build the app, or run scripts from your project files.
Every detected risk requires a concrete file, configuration, or store-detail signal and a clear confidence assessment.
Store-rule references point to Apple, Google, or Expo documentation and are reviewed when requirements change.
Usability, real runtime behavior, and the final installable app can require human or additional technical review.
The scanner identifies the framework, platforms, and relevant configuration sources before rules evaluate normalized facts such as permissions, purpose strings, dependencies, and listing claims. A finding is created only when a defined rule matches available evidence.
Each result should name the affected file, observed value, rationale, and prioritized action. Unresolved variables and generated native values are presented as uncertainty rather than proven defects.
Severity describes possible impact on review, security, or publication. Confidence describes how clearly the files support the conclusion. A critical signal with low confidence therefore needs verification first.
LaunchLint does not approve releases. The owner combines its prioritized risks with device tests, backend state, store-console settings, and the signed production artifact.
Rules prioritize official Apple, Google, Expo, Flutter, and Capacitor documentation. Sources, review dates, and affected rule areas are maintained editorially.
Material corrections receive a new expert-review date. The public editorial policy explains the process.
LaunchLint never installs dependencies, runs builds, or executes user code. This makes review bounded and reproducible but cannot prove runtime behavior, backend state, purchases, or store-account configuration.
AI may explain redacted scanner and policy evidence or turn it into tasks. It cannot create unsupported findings.
Supported app technologies
Complete means project structure, shipped dependencies, permissions, store configuration, and statically provable security patterns are assessed together.