prototype

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides structured guidance for creating throwaway prototypes. No malicious patterns, such as prompt injection, unauthorized data access, or remote code execution, were identified across the instructions and scripts.
  • [DATA_EXFILTRATION]: The skill includes explicit security boundaries by instructing the agent to avoid connecting prototypes to production databases and to favor in-memory state or isolated scratch files.
  • [INDIRECT_PROMPT_INJECTION]: While the skill describes UI components that react to URL parameters (?variant=) and terminal apps that process user keystrokes, these are standard input surfaces for the intended prototype functionality and are not presented in a way that creates a vulnerability.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 06:39 PM