The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user to run an unverified package from npm using npx --yes submit-expo-feedback@latest. Bypassing the confirmation prompt with --yes to execute code from a non-standard package name (not under the @expo scope) represents a potential risk where malicious code could be executed in the developer's environment.
[REMOTE_CODE_EXECUTION]: In references/new-architecture.md, the skill suggests using bunx xcobra to evaluate global variables in the running environment. xcobra is not a well-known or official Expo/React Native utility, posing a risk of executing unvetted code from the npm registry.
[EXTERNAL_DOWNLOADS]: The skill relies on fetching and executing several external packages at runtime without specifying versions or verifying their integrity, including submit-expo-feedback and xcobra.

expo-upgrade