The Agent Skills Directory

[SAFE]: The skill provides legitimate instructional content for internationalization and localization using well-known, industry-standard libraries and APIs. No malicious intent or hidden code was detected.
[SAFE]: Network operations mentioned in the documentation, such as TMS (Translation Management System) integration and CDN-level locale routing, are standard architectural components for global applications and do not involve unauthorized data exfiltration.
[SAFE]: The skill identifies a surface for Indirect Prompt Injection when using AI-powered translation pipelines to process untrusted source text. It mitigates this risk by recommending human review for safety-critical content and providing structural guidelines for translation consistency.
Ingestion points: Source code strings (JSX/HTML), user-supplied UI copy, and external translation files (SKILL.md).
Boundary markers: Recommends pseudo-localization with bracket wrapping to identify boundary issues (references/rtl-support.md).
Capability inventory: File system write operations for translation assets and network-based TMS synchronization (SKILL.md).
Sanitization: Emphasizes framework-level escaping and the use of ICU MessageFormat over string concatenation to prevent structural manipulation during translation (SKILL.md, references/icu-message-format.md).
[SAFE]: No hardcoded credentials, sensitive file access, or privilege escalation patterns were found. Placeholders used in documentation (e.g., for GitHub Action secrets) follow security best practices.

polyglot