The Agent Skills Directory

[SAFE]: The skill is restricted to reading local project files such as package.json, source code, and locale files to identify internationalization gaps. It does not perform network requests, write to the filesystem, or execute shell commands.
[INDIRECT_PROMPT_INJECTION]: While the skill processes untrusted input from codebase files (ingestion points: package.json, locale files, source code), it lacks dangerous capabilities (capability inventory: none) and includes a rule prohibiting file modifications, mitigating risks associated with malicious content in the audited files.

i18n