"Cerberus Guard team on Transforming Enterprise Through AI Hackathon"

Team Idea

Secure your production AI agents with CerberusGuard: a unified, defense-in-depth trust layer for prompt inspection, budget control, and safe execution

Team Members

Manuel Peña

manuel_pea615

Submission

CerberusGuard: Three-Head Defense for AI Agents

CerberusGuard unifies three MIT open-source primitives into defense in depth for AI agents: Lobster Trap (prompt inspection), PennyPrompt (budget control), and ClawCrate (execution sandbox), with one policy file and one correlated audit trail