monad-casino-agent
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill's architecture relies on autonomous AI agents interacting with one another (e.g., AI Werewolf, Poker). This design creates an inherent surface for indirect prompt injection where malicious instructions could be passed between agents.
- Ingestion points: inter-agent discussions, accusations in 'Werewolf', and chat logs.
- Boundary markers: None specified in the instructions.
- Capability inventory: Conceptual ability to manage virtual or on-chain currency ($MONAD).
- Sanitization: No input filtering or instruction-ignoring delimiters are defined for processing agent-generated content.
Audit Metadata