react-native-best-practices

Pass

Audited by Gen Agent Trust Hub on Oct 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and suggests the installation of several well-known third-party development tools and libraries (e.g., Spotify's Ruler, Shopify's FlashList, Re.Pack, and Flashlight). The documentation consistently includes security warnings to verify the provenance of these tools and avoid unsafe installation methods like piping remote scripts directly into a shell.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function involves analyzing external data such as profiler logs, bundle maps, and performance metrics. These ingestion points (found in js-profile-react.md and bundle-analyze-js.md) represent a theoretical attack surface where untrusted data could influence agent behavior. However, the skill mitigates this by providing structured workflows that focus on objective metrics and manual human verification.
  • [DYNAMIC_EXECUTION]: In references/bundle-code-splitting.md, the skill provides guidance on implementing remote JavaScript chunk loading. While dynamic execution of remote code is a high-risk capability, the skill includes a dedicated 'Security Model' section. It mandates the use of HTTPS-only first-party origins, signed release manifests, and code signing to ensure that only authorized release artifacts are executed.
  • [COMMAND_EXECUTION]: The skill utilizes platform-provided tools such as agent-device for profiling and standard development CLI tools (e.g., npx, gradlew, xcodebuild). All shell commands are scoped to legitimate performance analysis and build tasks, and the skill repeatedly reminds the user to review these operations before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 5, 2026, 09:27 PM