The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill manages potentially untrusted data from the browser (URLs, cookies, headers) used to determine active locales. It enforces security through a mandatory evidence chain: 1) Ingestion points include route segments in applyRouteLocale and detectLocale (in examples/lazy-loading.md and examples/ssr.ts); 2) Boundary markers are established via the isSupportedLocale validation; 3) Capability inventory shows no dangerous system or network operations (only local state changes and accessibility attributes); 4) Sanitization is performed by checking inputs against a hardcoded SUPPORTED_LOCALES whitelist.
[DYNAMIC_EXECUTION]: The skill demonstrates lazy-loading of locale files using dynamic import() calls with computed paths in examples/lazy-loading.md. This is handled securely by ensuring the locale variable is validated against a known-safe whitelist before the import is executed, preventing path traversal or unauthorized module loading.
[REMOTE_CODE_EXECUTION]: The skill references standard build-time optimization tools like @intlify/unplugin-vue-i18n and the vue-i18n library. All identified packages are well-known and reputable within the web development ecosystem, and no remote script piping patterns (such as curl | bash) are present.

web-i18n-vue-i18n