style-guide

Pass

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: LOWNO_CODE
Full Analysis
  • NO_CODE (SAFE): The skill consists entirely of Markdown files providing linguistic and formatting guidelines. There are no scripts (Python, JS, etc.), binaries, or system configurations present.
  • DATA_EXFILTRATION (SAFE): No network commands or data transfer patterns were found. URLs present in the example files (e.g., example.com/docs) are placeholders for documentation links within business templates.
  • PROMPT_INJECTION (SAFE): No malicious override instructions or jailbreak attempts were detected. Warning labels such as 'IMPORTANT: This document is confidential' are part of the mock business data and do not target the agent's control flow.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill provides the logic for correcting external text, which is a potential injection surface. However, because the skill lacks write, network, or execution capabilities, the risk is limited to potential misinformation in the generated output rather than system compromise.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 15, 2026, 08:36 PM