council

To run a multi-perspective debate, define your goal and context, spawn 3-5 agent teammates with diverse viewpoints, execute the structured debate protocol of framing, opening statements, and challenges, and receive a synthesized recommendation.

Multi-agent deliberation simulates a team of 3-5 agents with diverse viewpoints to discuss a decision, surface trade-offs, and identify robust solutions through a structured debate protocol ending in a synthesized recommendation.

Yes, multi-agent teams can be used for policy evaluation and design trade-offs. The deliberation process spawns agents with diverse viewpoints to discuss your specific context, improving decision quality through structured challenges and final synthesis.

The best way to structure deliberations among agent teams is using a four-phase protocol: framing, opening statements, challenges, and synthesis. This enforces disciplined debate across 3-5 agents to produce a synthesized recommendation.

You should avoid using a multi-perspective council when diverse viewpoints add no value to the decision. The deliberation process is specifically designed for architecture decisions, spec evaluations, or policy trade-offs where surfacing trade-offs is necessary.