translation-i18n

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill uses a structured multi-role core loop (WRITER, CRITIC, REVISER) and clear delimiters like <TRANSLATE_THIS>, <CONTEXT>, and <DOCUMENT> to isolate untrusted input from system instructions, effectively mitigating prompt injection risks.
  • [DATA_EXFILTRATION]: No unauthorized file access or data exfiltration patterns were detected. The skill's operations are restricted to processing localization files and performing authenticated web searches for glossary resolution.
  • [COMMAND_EXECUTION]: All file operations and format transformations are performed using deterministic parsing logic. No arbitrary command execution, shell spawning, or unsafe script evaluations were found.
  • [EXTERNAL_DOWNLOADS]: The skill includes a feature for web-based glossary resolution (L3 tier), which is managed through a dedicated tool abstraction and restricted by a document-level search cap to ensure efficiency and safety.
  • [SAFE]: The skill demonstrates security maturity through the implementation of HARD verification gates (M1/M2/M3) that deterministically check for placeholder integrity and glossary compliance before finalizing output.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 05:06 PM
Security Audit — agent-trust-hub — translation-i18n