Multi-agent orchestration with risk gates

Submit a goal. Agents reach consensus, score risk, auto-execute safe steps, and escalate dangerous ones for your sign-off.