translate

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, backdoors, or credential harvesting attempts were detected. The skill's operations align with its stated purpose of language translation.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface typical of translation tools that process untrusted external data.
  • Ingestion points: The skill reads user-provided text and local files (e.g., README.md) for translation, as defined in SKILL.md.
  • Boundary markers: The instructions lack specific delimiters to isolate the text being translated from the agent's instructions.
  • Capability inventory: The agent uses file read and write capabilities to process and save translated files.
  • Sanitization: No input validation or instruction filtering is performed on the source content before translation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 02:19 AM
Security Audit — agent-trust-hub — translate