m10-performance
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
- [Prompt Injection] (MEDIUM): The description field utilizes the prefix 'CRITICAL:', a known prompt injection pattern designed to hijack the agent's prioritization logic and override normal instruction processing.
- [Metadata Poisoning] (MEDIUM): The use of deceptive urgency markers in metadata attempts to influence the AI's internal reasoning about skill relevance and selection during runtime.
Audit Metadata