The Agent Skills Directory

[SAFE]: The skill provides instructions for standard internationalization (i18n) workflows in Elixir/Phoenix applications using the well-known gettext library.
[SAFE]: The documentation includes explicit security rules and code examples for validating user-controlled input (locale parameters) against a predefined allowlist. This implementation follows the principle of least privilege and prevents potential issues related to unvalidated input entering the system context.
[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data ingestion (locale from URL params/session) but provides robust sanitization logic.
Ingestion points: Locale data is read from conn.params["locale"] and get_session(conn, "locale") in the provided SetLocale plug.
Boundary markers: The instructions explicitly state to use an allowlist.
Capability inventory: No dangerous automated capabilities (network/file-write) are triggered directly by the input; it is used only to set the process locale via the Gettext backend.
Sanitization: The code snippet includes a validate function that checks the input against a hardcoded @supported_locales list.

gettext-i18n