humanizer

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFE
Full Analysis
  • [NO_CODE]: The skill consists entirely of Markdown instructions and documentation. There are no scripts (Python, Node.js, Shell, etc.) provided or executed by this skill.
  • [PROMPT_INJECTION]: The skill instructions define an 'editor' persona to humanize text. While it uses strong instructional language (e.g., 'You are a writing editor', 'Your Task'), these are legitimate configurations for the agent's behavior and do not attempt to bypass safety filters or extract system prompts.
  • [DATA_EXFILTRATION]: There are no commands or code patterns that attempt to access sensitive files or perform network operations. The allowed-tools section in SKILL.md restricts the environment, and the instructions themselves focus strictly on text transformation within the chat context.
  • [EXTERNAL_DOWNLOADS]: The README provides standard git clone instructions for manual installation into specific directories. The skill itself does not perform any automated downloads or package installations during operation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 02:40 PM