Multi-agent attack success evaluation