oma-translator

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the rg (ripgrep) utility to scan the local codebase for existing translations and terminology, ensuring consistent results without external risk.
  • [DATA_EXPOSURE]: Reads project-specific files including locale configurations (.json, .arb, .yaml), documentation, and source code to infer the appropriate register and domain context for the target language.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source text provided for translation. While this is a standard ingestion point, the instructions focus heavily on linguistic analysis, register matching, and mechanical verification, which naturally delimits the processing of potentially malicious instructions within the data.
  • [SAFE]: The skill includes robust verification gates (Stage 4) and humanization reviews (Stage 5-7) that focus on preventing 'translationese' and ensuring the integrity of placeholders and file structures.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 09:30 PM
Security Audit — agent-trust-hub — oma-translator