humanize-ai

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses a multi-layered logic framework for text processing. The operations described (體裁識別, 模式識別, 改寫執行) are standard prompt engineering techniques for high-quality text manipulation.
  • [PROMPT_INJECTION]: The skill processes untrusted user-provided text as part of its primary function, which presents an inherent surface for indirect prompt injection.
  • Ingestion points: User-provided text for 'AI-taste' diagnosis and rewriting in SKILL.md.
  • Boundary markers: The instructions do not define strict delimiters (such as XML tags or unique markers) to isolate the user text from the agent's internal instructions.
  • Capability inventory: The skill is restricted to text generation and analysis. There are no tools or instructions for network access, filesystem writes, or shell command execution defined.
  • Sanitization: There is no explicit sanitization or filtering logic applied to the input text prior to processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 01:36 AM
Security Audit — agent-trust-hub — humanize-ai